ARIADNE: Report on the ARIADNE Linked Data Cloud

D15.2: Report on the ARIADNE Linked Data Cloud

Authors: Franca Debole, CNR-ISTICarlo Meghini, CNR-ISTIGuntram Geser , SRFGDouglas Tudhope, USW

Ariadne is funded by the European Commission’s 7th Framework Programme.

Theviewsandopinionsexpressedinthisreportarethesoleresponsibilityoftheauthor(s)anddonotnecessarilyreflecttheviewsoftheEuropeanCommission.

ARIADNE–D15.2:ReportontheARIADNELinkedDataCloud,PreparedbyCNR-ISTI,SRFGandUSW(Public)

Version:1.0(final) 27thJanuary2017

Authors: FrancaDeboleandCarloMeghini,CNR-ISTI

GuntramGeser,SRFG

DouglasTudhope,USWContributingPartners

CeriBinding,USW

SaraDiGiorgio,MIBAC-ICCU

AchilleFelicetti,PIN

DimitrisGavrilis,ATHENARC

PhilippGerth,DAI

MariaTheodoridou,FORTH

Qualityreview: HollyWright,ADS

PaolaRonzino,PIN

ARIADNE–D15.2:ReportontheARIADNELinkedDataCloud PreparedbyCNR-ISTI,SRFGandUSW

ARIADNE 3 January2017

Tableofcontent

ExecutiveSummary.........................................................................................................................71 Introduction...............................................................................................................................82 Vision,studysummaries,andrecommendations......................................................................11

2.1 ArchaeologicalLinkedOpenData–avision.............................................................................112.2 Studysummariesandrecommendations.................................................................................12

2.2.1 LinkedOpenData:Backgroundandprinciples............................................................................122.2.2 TheLinkedOpenDataCloud........................................................................................................132.2.3 AdoptionoftheLinkedDataapproachinarchaeology...............................................................142.2.4 RequirementsforwideruptakeoftheLinkedDataapproach....................................................152.2.5 LinkedDatadevelopmentinARIADNE........................................................................................212.2.6 ARIADNELODCloud.....................................................................................................................22

3 LinkedOpenData:Backgroundandprinciples..........................................................................243.1 LOD–Abriefintroduction........................................................................................................243.2 Historicalandcurrentbackground...........................................................................................253.3 LinkedDataprinciplesandstandards.......................................................................................26

3.3.1 LinkedDatabasics........................................................................................................................263.3.2 LinkedOpenData.........................................................................................................................273.3.3 MetadataandvocabularyasLinkedData....................................................................................283.3.4 GoodpracticesforLinkedDatavocabularies...............................................................................293.3.5 MetadataforsetsofLinkedData.................................................................................................30

3.4 Whatadoptersshouldconsiderfirst........................................................................................313.5 MasteringtheLinkedDatalifecycle.........................................................................................323.6 Briefsummaryandrecommendations.....................................................................................33

4 TheLinkedOpenDataCloud.....................................................................................................354.1 LODCloudfigures.....................................................................................................................354.2 (Mis-)readingtheLODdiagram................................................................................................364.3 CulturalheritageintheLODCloud...........................................................................................384.4 Briefsummaryandrecommendations.....................................................................................41

5 AdoptionoftheLinkedDataapproachinarchaeology..............................................................435.1 Adoptionbyculturalheritageinstitutions................................................................................435.2 Lowuptakeforarchaeologicalresearchdata..........................................................................445.3 TheAncientWorldresearchcommunityasafront-runner.....................................................455.4 Briefsummaryandrecommendations.....................................................................................49

6 RequirementsforwideruptakeoftheLinkedDataapproach...................................................516.1 RaiseawarenessofLinkedData...............................................................................................51

6.1.1 Fragmentationofarchaeologicaldata.........................................................................................51



6.1.2 CurrentawarenessofLinkedData...............................................................................................526.1.3 Briefsummaryandrecommendations........................................................................................54

6.2 ClarifythebenefitsandcostsofLinkedData...........................................................................556.2.1 Thenotionofanunfavourablecost/benefitratio.......................................................................556.2.2 Lackofcost/benefitevaluation....................................................................................................566.2.3 Collectingexamplesofbenefitsandcosts...................................................................................586.2.4 Briefsummaryandrecommendations........................................................................................62

6.3 Enablenon-ITexpertsuseLinkedDatatools...........................................................................636.3.1 LinkedDatatools:therearemanyandmostarenotuseable.....................................................636.3.2 Needofexpertsupport................................................................................................................646.3.3 ThecaseofCIDOCCRM:fromdifficulttodoable........................................................................646.3.4 Progressthroughdatamappingtoolsandtemplates..................................................................656.3.5 Needtointegratesharedvocabulariesintodatarecordingtools...............................................666.3.6 Briefsummaryandrecommendations........................................................................................68

6.4 PromoteKnowledgeOrganizationSystemsasLinkedOpenData...........................................696.4.1 KnowledgeOrganizationSystems(KOSs)....................................................................................696.4.2 Culturalheritagevocabulariesinuse...........................................................................................706.4.3 DevelopmentofKOSsasLinkedOpenData................................................................................716.4.4 KOSsregistries.............................................................................................................................746.4.5 Briefsummaryandrecommendations........................................................................................76

6.5 FosterreliableLinkedDataforinterlinking..............................................................................776.5.1 Currentlackofinterlinking..........................................................................................................776.5.2 Whyistherealackofinterlinking?..............................................................................................786.5.3 NeedofreliableLinkedDataresources.......................................................................................786.5.4 FosteracommunityofarchaeologicalLODcurators...................................................................806.5.5 Briefsummaryandrecommendations........................................................................................80

6.6 PromoteLinkedOpenDataforresearch..................................................................................816.6.1 ALinkedOpenDatavision(2010)................................................................................................826.6.2 LODforresearch:Thecurrentstateofplay.................................................................................826.6.3 Searchvs.research......................................................................................................................846.6.4 Examplesofresearch-orientedLinkedDataprojects..................................................................856.6.5 CIDOCCRMasabasisforresearchapplications..........................................................................866.6.6 Briefsummaryandrecommendations........................................................................................88

7 LinkedDatadevelopmentinARIADNE......................................................................................897.1 TheARIADNEcatalogueasLinkedOpenData..........................................................................897.2 WorkonvocabulariesasLinkedData.......................................................................................90

7.2.1 VocabulariesinSKOS...................................................................................................................907.2.2 Mappingofsubjectvocabularies.................................................................................................927.2.3 MetadataforvocabulariesandmappingsinSKOS......................................................................94

7.3 What–Where–WhenasLinkedData.....................................................................................947.3.1 What(subjects)............................................................................................................................94



7.3.2 Where(places).............................................................................................................................957.3.3 When(chronology)......................................................................................................................95

7.4 UseofvocabulariesinNLPanddatamining............................................................................967.4.1 NaturalLanguageProcessing.......................................................................................................967.4.2 MiningofLinkedData..................................................................................................................97

7.5 CIDOCCRMextensionsandmappings.....................................................................................997.6 DemonstratorsusingCRM-basedLinkedData.......................................................................1017.7 Briefsummaryandlessonslearned.......................................................................................104

8 ARIADNELODCloud...............................................................................................................1068.1 TheARIADNELODCloud–inbrief.........................................................................................1068.2 Architecture............................................................................................................................1078.3 TheLinkedOpenDataServer.................................................................................................1088.4 TheDemonstrators.................................................................................................................1128.5 TheMappingandOntologyServer.........................................................................................1138.6 Promotionofexternaluse......................................................................................................1158.7 Briefsummaryandlessonslearned.......................................................................................116

9 Referencesandrelevantothersources...................................................................................118



AcronymsofARIADNEpartners

AIAC AssociazioneInternazionalediArcheologiaClassica(Italy)

ARHEO ArheovestTimisoaraAssociation(Romania)

ARUP-CAS ArcheologickyustavAVCR,Praha,v.v.i./InstituteofArchaeologyoftheAcademyofSciences(CzechRepublic)

Athena-DCU AthenaResearchandInnovationCenterinInformationCommunicationandKnowledgeTechnologies/DigitalCurationUnit(Greece)

CNR ConsiglioNazionaledelleRicercheinstitutes,CNR-ISTIandCNR-ITABC(Italy)

CSIC-Incipit ConsejoSuperiordeInvestigacionesCientificas/SpanishNationalResearchCouncil,InstituteofHeritageSciences(Spain)

CYI-STARC TheCyprusInstitute,ScienceandTechnologyinArchaeologyResearchCenter

DAI DeutschesArchäologischesInstitut(Germany)

Discovery TheDiscoveryProgrammeLBG(Ireland)

FORTH-ICS FoundationforResearchandTechnologyHellas,InstituteofComputerScience(Greece)

INRAP InstitutNationaldesRecherchesArchéologiquesPréventives(France)

KNAW-DANS NetherlandsAcademyofArtsandSciences,DataArchivingandNetworkedServices(Netherlands)

LeidenU LeidenUniversity,FacultyofArchaeology(Netherlands)

MiBAC-ICCU ItalianMinistryofCulturalAssetsandActivities-CentralInstitutefortheUnionCatalogue(Italy)

MNM-NOK MagyarNemzetiMúzeum,NemzetiÖrökségvédelmiKözpont/HungarianNationalMuseum,NationalHeritageProtectionCentre(Hungary)

NIAM-BAS NationalInstituteofArchaeologywithMuseumoftheBulgarianAcademyofSciences(Bulgaria)

ÖAW-OREA ÖsterreichischeAkademiederWissenschaften,InstitutfürOrientalischeundEuropäischeArchäologie(Austria)

PIN PIN-ServiziDidatticieScientificiperl’UniversitàdiFirenzes.c.r.l.(Italy)

SND SwedishNationalDataService(Sweden)

SRFG SalzburgResearchForschungsgesellschaftm.b.H.(Austria)

USW UniversityofSouthWales(UnitedKingdom)

ADS-UoY ArchaeologyDataService,UniversityofYork(UnitedKingdom)

ZRC-SAZU ScientificResearchCentreoftheSlovenianAcademyofSciencesandArts,InstituteofArchaeology(Slovenia)



ExecutiveSummary

This report has been producedwithin the ARIADNE project as part ofWork Package 15, “LinkingArchaeological Data”. This document is a deliverable (D15.3) of the ARIADNE project (“AdvancedResearch Infrastructure forArchaeologicalDatasetNetworking in Europe”),which is fundedunderthe European Community's Seventh Framework Programme. It presents the results of the workcarriedout inTask15.3“ARIADNELinkedDataCloud”.TheoverallobjectiveofARIADNE is tohelpmakingarchaeologicaldatabetterdiscoverable,accessibleandre-useable.TheprojectaddressesthefragmentationofarchaeologicaldatainEuropeandpromotesacultureofopensharingand(re-)useof data across institutional, national and disciplinary boundaries of archaeological research.Morespecifically,ARIADNEimplementsane-infrastructurefordatainteroperability,sharingandintegratedaccessviaadataportal.LinkedOpenDatacangreatlycontributetothesegoals.

Lessonslearned,recommendationsandbriefconclusionsareincludedattheendofeverysection.



1 Introduction

TowardsawebofarchaeologicalLinkedOpenData–avision

TheARIADNELinkedOpenData“cloud”isenvisionedasawebofsemanticallyinterlinkedresourcesof and for archaeological research. Archaeology is amulti-disciplinary field of research, hence theweb of Linked Data initiated by different projects, including ARIADNE, spans data resources ofvariousdomains and specialties, for examplehistory and geographyof the ancientworld, classics,medieval studies, cultural anthropology and various data from the application of natural sciencemethodstoarchaeologicalresearchquestions(e.g.physical,chemicalandbiologicalsciences).

Oneof themain objectives of theARIADNEproject has been to provide the archaeological sectorwithadatainfrastructureandportalfordiscoveringandaccessingdatasetswhicharebeingsharedby research institutions and digital archives located in different European countries. Theinfrastructureandportalarenotstand-aloneimplementationsbutserveasanodeintheecosystemof e-infrastructure services for archaeology and various related disciplines, including otherhumanities as well as social, natural, environmental and life sciences. To become such a node,interoperabilitywithexternalservicesisrequiredandcanbeimplementedbasedontheLinkedDataapproach.

LinkedDatasupportinARIADNE

WP15supportsthedevelopmentofLinkedOpenDatawithinandbeyondtheproject.Theactivitiesofthisstrandofworkconcerned:

o themetadataofthedatasetsregisteredintheARIADNEdatacatalogue,

o vocabularies for the metadata describing registered datasets (e.g. mapping of existingvocabularies,supportforthegenerationofvocabulariesinSKOS),

o mappingofdatasetstothecoreCIDOCCRMandextensionsoftheCRMcreatedinARIADNE,

o demonstrators generating and using Linked Data (e.g. metadata extracted from unstructureddatasuchasgreyliterature,explorationofCIDOCCRMbaseddata),and

o providingaccesstoARIADNELinkedDataforexternalapplicationdevelopers.

Thus the work centred on Linked Data related to data registration, enabling data integration viavocabularies and the CIDOC CRM ontology, demonstration of enhanced or new capabilities, andmakingtheARIADNEdatacatalogueandotherresultsoftheseactivitiesaccessiblethroughagraphdatabaseor“cloud”ofLinkedData.

CurrentlevelofLODadoptioninarchaeology

Thelast10yearshaveseensubstantialprogress inLODexpertise, i.e.what isrequiredtoproduce,publishand interlinkLODfromculturalheritagecollections (e.g.museumartefactcollections).Thisexpertisehasbeenacquiredmostlythroughexperimentalprojects,andonlyafewculturalheritagedatasetsareeffectivelyinterlinkedasyet.Withregardtoarchaeologicaldataspecifically,fewLinkedDatadatasetshavebeenproducedandhardlyanyshowuponthewell-knownLODClouddiagram.Incomingyearsamuchwideruptakeof theLODapproach in thedomain isnecessary, so thata richwebofdatacanemerge.



Requirementsforawideruptake

WP15 activities took into account factors that currently impede the development of a web ofsemantically interlinked archaeological data. Therefore the present report particularly addressesrequirements for a wider uptake of a Linked Data approach in archaeology. The study of theserequirementswillbevaluableformanywhohavetakenaninterestinLinkedOpenData(LOD),wouldlikeanoverviewofthecurrentsituationinculturalheritageandarchaeology,andrecommendationsonhowtoadvancetheavailabilityandinterlinkingofLODinthisfield.

Specificactionsarerecommendedto:

o raiseawarenessofLinkedData,

o clarifythebenefitsandcostsofLinkedData,

o enablenon-ITexpertsuseLinkedDatatools,

o promoteKnowledgeOrganizationSystemsasLinkedOpenData,

o fosterreliableLinkedDataforinterlinking,

o promoteLinkedOpenDataforresearch.

Amongthevariousrequirements,theimportanceoffosteringacommunityofLODcuratorswhotakecareforpropergeneration,publicationandinterlinkingofarchaeologicaldatasetsandvocabularieswerehighlighted.

LessonslearnedinthedevelopmentofLODwithinARIADNE

Onefindingisthecriticalimportanceofthesubjectvocabularies,e.g.theGettyArtandArchitectureThesaurus(AAT),combinedwiththeCIDOCCRMontologyentities,whichactaslinkinghubsforthewebofdata.ThisisthemostobviousroutetoconnectionwithexternalLOD.Moreworkisneededon the identification of further linking hubs, for example the Period0 set of cultural periods. Themappingofdatasets tosuchhubsrequiresdomainknowledge,easy touse tools,andguidance foruserswhoarecarryingoutsuchworkforthefirsttime.Whilerecommendedtoolsarehelpful,fullyautomatedmapping appears unlikely to achive quality results at the current time. There ismuchscopetoexploretheutilityofLODinpractice,takingaccountoftheobjectivesandrequirementsofdifferent user communities. There is still a way to go before advanced uses of LOD will becomeapplicableandbeneficialinonlineresearchenvironments;moreeffortmustbeinvestedtomakethishappen. In order to motivate user organisations to work with Linked Data, exemplar workingapplicationsareneededthataddressarealuser(scientific/research)need.SuchexemplarsmightbeenduserapplicationsorprogrammaticinterfacestotheunderlyingLOD.

BuildingtheARIADNELODCloud–lessonslearned

WhiletheLinkedOpenDatastandardsareessentialforintegratingdata,thetechnologysupportingsuchintegrationisstillinitsinfancy.TheARIADNELOD,comprisedofLODderivedfromtheARIADNEcatalogue, isrepresentedbythreedemonstratorsandvariousvocabularies,andhasresultedinthecreationofabout32millionRDFtriples.Whileanyrelationaldatabasecaneasilyhandlemillionsofrecords, the corresponding volume of RDF in a current triple store can cause serious efficiencyproblemsasexperiencedintheexperimentationwiththeARIADNELinkedDataCloud,andthatthisisthepricetobepaidforinteroperability.Morerobustandefficientgraphdatabasesarerequiredifwewant to proceed towards Big Data as LinkedData. This is the firstmajor lesson learnedwhileimplementingtheARIADNELinkedDataCloud.



Thesecondlessoncomesfromthegraphdatamodel.Thismodelisintrinsicallybinary,whichmakesitdifficulttoexpresshigherrankrelations,andtoeasilyimplementdataconnectionpatterns.Inthelatter case, the patterns may involve data chains that span several arcs, and their definition andimplementationisnottrivial.Conversely,correlationsbetweendataitemscanbeepitomizedbysuchpaths,whichneedtobedetected,andthis isacomputationallyvery intensivetask if the lengthofthepathsgobeyond2-3arcs.Thisfacthasalwaysbeenknownfromatheoreticalpointofview,butworkingwithrealdatawecouldexperienceitinpractice.



2 Vision,studysummaries,andrecommendations

Thischaptersummarisestheresearchanddevelopmentresultspresentedinthisreport.IthighlightsavisionofawebofarchaeologicalLinkedOpenData(LOD),addressestheLODprinciplesandwebofLinked Data (the “LOD Cloud”), the adoption of the LOD approach so far in archaeology, andrequirements for a wider uptake in the sector. Moreover the chapter summarises the LODdevelopment inARIADNEandhowthegenerateddata isbeingmadeavailablebeyondtheproject.ThesectionsalsoproviderecommendationsonhowtoincreasetheadoptionoftheLODapproachinarchaeologyandlessonslearnedintheworkonLODintheARIADNEproject.

2.1 ArchaeologicalLinkedOpenData–avision

This report envisions the emergence of a web of semantically interlinked resources of and forarchaeological research based on the Linked Data approach. Over the next 5-10 years a web ofLinkedOpenDatacouldbebuiltthatspansvocabulariesanddataofarchaeological,culturalheritageandrelatedfieldsofresearch.

About10yearsago therewere considerabledoubtsabout theuptakeof SemanticWeb standardsand technologies. Reasons for this doubtwere centred on the still on-going standardisationwork,little experience of implementation under real world conditions, and expected high costs ofconversionoflegacymetadataandknowledgeorganizationsystems(e.g.thesauri)toSemanticWebstandards.

In recent years the Linked Data approach has seen substantial progress with regard to maturestandards,availableexpertiseand tools,andexamplesofdatapublicationand linking.RecognitionanduptakeoftheapproachhasgrownfarbeyondtheinitiallysmallpioneeringgroupsofLinkedDatadevelopers. The Open Data movement has been an important driver for this development,particularly through the involvement of governmental and public sector agencies, who havepromotedstandardsandimplementeddatacataloguesandportals.

TheLinkedDataapproachhasbeenembracedbyseveral researchcommunities, forexample,geo-spatial, environmental and some natural sciences (e.g. bio-sciences). Also the cultural heritagesector, particularly the library and museum domains, have been among the early adopters. Thusthere is already potential for interlinking and enriching archaeological research data with specificinformation,aswellaswithinawidercontext.

Archaeology is a multi-disciplinary field of research, hence the web of Linked Open Data couldinclude resources of various domains and specialties, for example history and geography of theancientworld,classics,medievalstudies,culturalanthropologyandvariousdatafromtheapplicationof natural sciences methods to archaeological research questions (e.g. physical, chemical andbiological sciences). Also data of geo-spatial, environmental and earth sciences are relevant toseveralfieldsofarchaeologicalresearch.

Butwideanddeepinterlinkingwillrequirerichintegrationofconceptualknowledge(ontologies)andterminologies from different domains. Integration could be progressed based on use caseswith aclearaddedvalueforarchaeologicalandotherresearchcommunities.Suchusecaseswouldsupportinterdisciplinary research involving researchers in archaeology and other domains, natural historyandenvironmentalchange,forinstance.

Asamulti-disciplinaryareaofresearch,archaeologycouldbenefitgreatlyfromacomprehensivewebofLinkedOpenData,involvingdataandvocabulariesofallrelateddisciplines.However,firstthereis



stillalotofhomeworktodobyresearchinstitutions,projectsandarchivessothatanarchaeologicalwebofLinkedOpenDatawillemergeandbecomeinterlinkedwithresourcesofotherdisciplinesaswellasrelevantpublicsectorinformation.

2.2 Studysummariesandrecommendations

2.2.1 LinkedOpenData:Backgroundandprinciples

Briefsummary

The term LinkedData refers to principles, standards and tools for the generation, publication andand linkingof structureddatabasedon theW3CResourceDescription Framework (RDF) familyofspecifications.

ThebasicconceptofLinkedDatawasdefinedbyTimBerners-Leeinanarticlepublishedin2006.Thisconcept helped to re-orientate and channel the initial grand vision of the Semantic Web into aproductive new avenue. Previously the research and development community presented theSemanticWeb vision as a complex stack of standards and technologies. This stack seemedalways“under construction” and together with the difficult to comprehend Semantic Web terminology,createdtheimpressionofanacademicactivitywithlittlerealworldimpact.

In 2010 Berners-Lee’s request for Linked Open Data aligned Linked Data with the Open Datamovement.Sincethen,thequestforLinkedOpenData(LOD)hasbecomeparticularlystronginthegovernmental/publicsectoraswellasinitiativesforculturalandscientificLOD.

LinkedDataprinciplesincludethatadatapublishershouldmakethedataresourcesaccessibleontheWeb viaHTTPURIs (UniformResource Identifiers),which uniquely identify the resources, anduseRDFtospecifypropertiesofresourcesandofrelationsbetweenresources.InordertobeLinkedDataproper, thepublishersshouldalso linktoURI-identifiedresourcesofotherproviders,henceaddtothe“webofdata”andenableuserstodiscoverrelatedinformation.AndtobeLinkedOpenDatathepublishermustprovidethedataunderanopen license(e.g.CreativeCommonsAttribution[CC-BY]orreleaseitintothePublicDomain).

The Linked Data approach allows opening up “data silos” to the Web, interlinking of otherwiseisolated data resources, and enables re-use of the interoperable data for various purposes. Thelandscapeofarchaeologicaldata ishighly fragmented.ThereforeLinkedDataareseenasaway tointerlink dispersed and heterogeneous archaeological data and, based on the interlinking, enablediscovery,accesstoandre-useofthedata.

Building semantic e-infrastructure and services for a specific domain such as archaeology requirescooperation between domain data producers/curators, aggregators and service providers.Cooperation is necessary not only for sharing datasets through a domain portal (i.e. theARIADNEdataportal),butalsotousecommonoralignedvocabularies(e.g.ontologies,thesauri)fordescribingthedatasothatitbecomesinteroperable.

In addition to the basic Linked Data principles there are also specific recommendations forvocabularies. Particularly important is re-using or extending wherever possible establishedvocabulariesbeforecreatinganewone.The rationale for re-use is thatdifferent resourceson theweb of Linked Data which are described with the same or mapped vocabulary terms becomeinterlinked. This makes it easier for applications to identify, process and integrate Linked Data.Moreover,re-useandextensionofexistingvocabulariescanlowervocabularydevelopmentcosts.



ItisalsorecommendedtoprovidemetadataforLinkedDataofdatasetsaswellasvocabularies.TheVocabulary of Interlinked Datasets (VoiD) is often being used to provide suchmetadata. It is alsogoodpractice to register sets of LinkedData in a domain data catalogue and/or general registriessuchas theDataHub.Furthermore thepublisher shouldannounce thedatasetvia relevantmailinglists,newslettersetc.andinviteotherstoconsiderlinkingtothedataset.

Linked Data should not be published “just in case”. Rather publishers should consider the re-usepotential and intended or possible users of their data. As Linked Data consumers they need toaddress the question of which data of others they could link to. These questionsmake clear theimportance of joint initiatives for providing and interlinking datasets of certain domains such asarchaeology.

Recommendations

o UsetheLinkedDataapproachtogeneratesemanticallyenhancedandlinkedarchaeologicaldataresources.

o Participate in joint initiatives for providing and interlinking archaeological datasets as LinkedOpenData.

o Choose datasets which allow generating value if made openly available as Linked Data andconnectedwithotherdata,includinglinkingofthedatasetsbyothers.

o Re-useexistingLinkedDatavocabularieswhereverpossibleinordertoenableinteroperability.

o Describe the Linked Data with metadata, including provenance, licensing, technical and otherdescriptiveinformation.

o Registerthedatasetinadomaindatacatalogueand/orgeneralregistriessuchastheDataHub.Alsoannouncethedatasetviarelevantmailinglists,newslettersetc.andinviteotherstoconsiderlinkingtothedataset.

2.2.2 TheLinkedOpenDataCloud

Briefsummary

TheLinkedOpenDataCloud is formedbydatasets thatareopenlyavailableontheWeb inLinkedDataformatsandcontainlinkspointingatothersuchdatasets.OnetaskoftheARIADNEprojectistopromote the emergence of a web of interlinked archaeological datasets which comply with theLinkedOpenData(LOD)principles.ItisanticipatedthatthiswebofarchaeologicalLODwillbecomepartofthewiderLODCloudandinterlinkedwithrelatedotherdataresources.

ThelatestLODClouddiagram(2014)includesonlyfewsetsofculturalheritageLODandtheydonotform a closely linkedweb of LinkedData. None of the datasets concerns archaeology specifically.AdditionalsetsofculturalheritageLinkedDataexist,afewofwhicharearchaeological,butin2014they did not conform to the criteria for being included in the LOD Cloud diagram (e.g. therequirementofbeingconnectedviaRDFlinkswithatleastoneothercompliantdataset).

MaybethenextversionoftheLODClouddiagramwillcontainsomeoftheearlierandmorerecentsetsofarchaeologicalLinkedOpenData.Hopefullythiswillincludesomerelevantvocabularieswhichrecentlyhavebeen transformed to LinkedData in SKOS format. In 2014 theonly cultural heritagevocabularyonthediagramwastheArt&ArchitectureThesaurus(AAT),whichhasthepotential tobecomeoneofthecorelinkinghubsforculturalheritageinformationintheLODCloud.The LOD Cloud is not a single entity but represents datasets of different providers that aremadeavailable indifferentways(e.g.LDserver,SPARQLendpoint,RDFdump)andtheresourcesmaybe



unreliable, e.g. some SPARQL endpoints are off-line. There is no centralmanagement and qualitycontrol of the LODCloud.Webs of reliable and richly interlinked datasets are only presentwherethere isacommunityofLinkedDataproducersandcurators(e.g. intheareasofbio-medical& lifesciencesorlibraries).Cultural heritage is not yet an area of densly interlinked and reliable LOD resources; so far acommunity of cooperating LOD producers and curators has not solidified. Targeted activities tofoster and support further publication and interlinking of datasets are required so that a web ofarchaeological, cultural heritage and other relevant datawill becomemore establishedwithin theoverallLinkedOpenDataCloud.

Recommendations

o Encourage more archaeological institutions and repositories to publish the metadata of theirdatasets (collections,databases)asLinkedOpenData;alsopromotepublicationofdomainandproprietaryvocabulariesofinstitutionsasLOD.

o Foster the formation of a community of archaeological LOD producers and curators whogenerate,publishandinterlinkLOD,includinglinking/mappingbetweenvocabularies.

2.2.3 AdoptionoftheLinkedDataapproachinarchaeology

Briefsummary

Intheareasaddressedbythisstudy,culturalheritageinstitutionsareamongtheleadingadoptersoftheLinkedDataapproach.TheAncientWorldandClassicsresearchcommunity isafront-runnerofuptake on the research side, while there have been only few projects around Linked Data usingarchaeologicalresearchdata.

Thissituationisduetoconsiderabledifferencesbetweenculturalheritageinstitutionsandresearchprojects, and between projects in different domains of research. For cultural heritage institutionssuch as a libraries, archives andmuseums adoption of LinkedData is in linewith theirmission tomake information about heritage readily available and relevant to different user groups, includingresearchers.AdoptionhasalsobeenpromotedbyinitiativessuchasLOD-LAM,theInternationalLODin Libraries, Archives, and Museums Summit (since 2011). In the field of archaeological researchtherewere no such initiatives or only at small scale, for example sessions at CAA conferences ornational thematic workshops. But promotional activities, particularly at the national level, areimportanttoreacharchaeologicalinstitutesandresearchgroupsandmakethemawareoftheLinkedDataapproach.

AdoptionintheAncientWorldandClassicsresearchcommunity isbeingdrivenbyspecialitiessuchasnumismaticsandepigraphy,wherethereareinitiativestoestablishcommondescriptivestandardsbasedonLinkedDataprinciples.Thegoalistoenableannotationandinterlinkingofinformationofspecialcollectionsorcorporaforresearchpurposes.Thiscommunityhasledthewaybyfocussingoncertaintypesofartefacts (inscriptions,coins,ceramicsandothers),whichprovideclearadvantageswithregardtotheeaseofusingtheLinkedDataapproach.

AgooddealoftherecognitionoftheAncientWorldandClassicsresearchcommunitybeingafront-runner inLinkedDatastemsfromthePelagios initiative.Pelagiosprovidesacommonplatformandtools forannotatingandconnectingvarious textual resources (both theclassical textandscholarlyreferences) based on place references. Pelagios clearly demonstrates benefits of contributing andassociatingdataderivedfromdifferentcontributorsbasedonalight-weightLinkedDataapproach.



The data generated by the myriad forms of Archaeological fieldwork present a more difficultsituation,inthatabasicunitofresearchcanbeasiteoranentirelandscape,wherearchaeologistsmay document a variety of structures, cultural remains, artefacts and biological material, using avarietyofmethods.Theheterogeneityofthearchaeologicaldataandthe“site”asafocusofanalysispresentsasituationwherethebenefitsofLinkedData,whichwouldrequiresemanticannotationofthevarietyofdifferentdatawithcommonvocabularies,arelessapparent.ThereforeadoptionoftheLinkedDataapproachcanbehardly foundatthe levelof individualarchaeologicalexcavationsandother fieldwork, but, in a few cases, community-level data repositories and databases of researchinstitutes.Repositoriesanddatabases,notindividualprojects,shouldalsoinnextyearsbetheprimetargetwhenpromotingtheLinkedDataapproach.

Allproponentsof theLinkedDataapproach, including theARIADNELinkedDataSIGaswellas thedirectorsofthePelagiosinitiative,agreethatmuchmoreneedstobedonetoraiseawarenessoftheapproach,promoteuptake,andprovidepracticalguidanceandeasytousetoolsforthegeneration,publicationandinterlinkingofLinkedData.

Recommendations

o MoreneedstobedonetoraiseawarenessandpromoteuptakeoftheLinkedDataapproachforarchaeological researchdata. Inaddition to sessionsat international conferences, promote theapproachtostakeholderssuchasarchaeologicalinstitutesatthenationallevel.

o The prime target when promoting the approach should be persistent data repositories anddatabasesofresearchinstitutes(notindividualprojects).

o To drive uptake provision of practical guidance and easy to use tools for the generation,publicationandinterlinkingofLinkedDataisnecessary.

o Promotetheuseofestablishedandemergingsemanticdescriptionandannotationstandardsforartefactssuchascoins,inscriptions,ceramicsandothers;forbiologicalremainsofplants,animalsand humans suggest using available relevant biological vocabularies (e.g. authoritative speciestaxons,lifescienceontologies,andothers).

o ContributetothePelagiosplatform(whereappropriate)oraimtoestablishsimilarhigh-visibilitydatalinkingprojectsforarchaeologicalresearchdata.

2.2.4 RequirementsforwideruptakeoftheLinkedDataapproach

RaiseawarenessofLinkedData

Briefsummary

LinkedDataenablesinteroperabilityofdispersedandheterogeneousinformationresources,allowingthe resources to become more discoverable, accessible and re-useable. In the fragmented datalandscapeof archaeology this is substantial task. In theARIADNEonline survey, in addition to theexpectationsof thearchaeological researchcommunityaround thecreationofadataportal,werecross-searching of data archives with innovative, more powerful search mechanisms. But suchexpectationswerenotnecessarilyassociatedwithcapabilitiesofferedbyLinkedData.Thereforethegapbetweenadvantagesexpectedfromadvancedservicesand“buyin”andsupportoftheresearchcommunityforLinkedDatamustbeclosedbytargetedactions.

A small survey of the AthenaPlus project (2013) indicated that cultural heritage organisations arealready aware of Linked Data, but few had first-hand experience with such data. Among theexpectations from connecting their own and external Linked Data resources, was increasing the



visibility of collections and creating relations with various other information resources. Somerespondents also considered possible disadvantages, e.g. loss of control over their own data or adecreaseindataqualityduetolinkstonon-authoritativesources.

IntheARIADNEonlinesurvey(2013)“Improvements in linkeddata”, i.e. interlinkingof informationbasedonLinkedDatamethodstoenablebetter informationservices,wasconsideredmorehelpfulby repository managers than researchers. Researchers perceived interlinking of information asimportant,butmaynotseethisasanareafortheirownresearch.Indeed,individualresearchersandresearchgroupsshouldmaynotbethoughtofasaprimaryfocusofLinkedDatainitiatives.Managersofdigitalarchivesfortheresearchcommunityandinstitutionalrepositoriesaremuchmorerelevanttargetgroups.Furthermoredatamanagersoflargeandlong-termarchaeologicalprojectsshouldbeaddressedastheywillalsoconsiderrequiredstandardsfordatamanagementandinterlinkingmorethoroughly.

Recommendations

o AddressthehighlyfragmentedlandscapeofarchaeologicaldataandhighlightthatLinkedDatacanallowdispersedandheterogeneousdataresourcesbecomebetterintegratedandaccessible.

o Consider as primary target group of Linked Data initiatives not individual researchers butmanagersofdigitalarchivesandinstitutionalrepositories.

o Includealsodatamanagersand IT staffof largeand long-termarchaeologicalprojectsas theywillalsoconsiderrequiredstandardsfordatamanagementandinterlinkingmorethoroughly.

ClarifythebenefitsandcostsofLinkedData

Briefsummary

There isawidespreadnotionofanunfavourable ratioofcostscompared tobenefitsofemployingSemanticWeb / Linked Data standards for informationmanagement, publication and integration.This notion should be removed as it is a strong barrier to a wider adoption of the Linked Dataapproach.

The basic assumption of Linked Data is that the usefulness and value of data increases themorereadily it can combined with relevant other data. Convincing tangible benefits of Linked Datamaterialiseifinformationproviderscandrawonownandexternaldataforenrichingservices.Thereareexamplesforsuchbenefits,e.g.inthemuseumcontext,butnotyetforarchaeologicalresearchdata. Importantly, intherealmofresearchbenefitsofLinkedDataare lessaboutenhancedsearchservices but research dividends, e.g. discovery of interesting relations or contradictions betweendata.

LinkedDataprojectstypicallymentionsomebenefits(e.g.integrationofheterogeneouscollections,enrichedinformationservices),butverylittleisknownaboutthecostsofdifferentprojects.Thereisaclearneedtodocumentanumberofreferenceexamples,forexample,whatdoesitcosttoconnectdatasetsviasharedvocabulariesor integratedatabasesthroughmappingthemtoCIDOCCRM,andhowdoesthatcomparetoperceivedbenefits?AlthoughvocabulariesplayakeyroleinLinkedDataastonishinglittleisalsoknownaboutthecostsofemployingvariousKOSs.

Somemethodsandtoolsappear tohavereducedthecostofLinkedDatagenerationconsiderably,OpenRefineormethodstooutputdatainRDFfromrelationaldatabases,forinstance.AsthereisaproliferationoftoolspotentialLinkedDataprovidersneedexpertadviceonwhattouse(andhowto



use it) for their purposes and specific datasets, taking account alsoof existing legacy systems andstandardsinuse.

Recommendations

o Proponents of the Linked Data approach should address the widespread notion of anunfavourable ratio of costs compared to benefits of employing Semantic Web / Linked Datastandards.

o Major benefits of Linked Data can be gained from integration of heterogeneous collections/databasesandenhancedservicesthroughcombiningownandexternaldata.Butexamplesthatclearlydemonstratesuchbenefitsforarchaeologicaldataareneeded.

o Inordertoevaluatethecosts,informationaboutthecostfactorsanddriversshouldbecollectedandanalysed.AgoodunderstandingofthecostsofdifferentLinkedDataprojectswillhelpreducethecosts,forexamplebyprovidingdedicatedtools,guidanceandsupportforcertaintasks.

o MoreinformationwouldbewelcomeonhowspecificmethodsandtoolshaveallowedinstitutionsreducingthecostsofLinkedDatainprojectsofdifferenttypesandsizes.

o Generalrequirementsforprogressaremoredomain-specificguidanceandreferenceexamplesofgoodpractice.

Enablenon-ITexpertsuseLinkedDatatools

Briefsummary

Showcase examples of Linked Data applications in the field of cultural heritage (e.g. museumcollections)sofardependedheavilyonthesupportofexpertswhoarefamiliarwiththeLinkedDatamethods and required tools (often their own tools). But such know-how and support is notnecessarilyavailableforthemanyculturalheritageandarchaeologyinstitutionsandprojectsacrossEurope.AmuchwideruptakeofLinkedDatawillrequireapproachesthatallownon-ITexperts(e.g.subjectexperts,curatorsofcollections,projectdatamanagers)domostoftheworkwitheasytousetoolsandlittletrainingeffort.

Anumberofprojectshavereportedadvancesinthisdirectionbasedontheprovisionofusefuldatamapping recipes and templates, proven tools, and guidance material. For example, the STELLARLinkedData toolkithasbeenemployed in severalprojectsandappears tobeuseablealsobynon-expertswithlittletrainingandadditionaladvice.

Good tutorials and documentation of projects are helpful, but the need for expert guidance invariousmattersofLinkedOpenDataisunlikelytogoaway.Forexample,therearealotofimmature,nottriedandtestedsoftwaretoolsaround.Thereforeadviceofexpertsisnecessaryonwhichtoolsare reallyprovenandeffective for certain tasks, andprovidersof such tools shouldofferpracticaltutorialsandhands-ontraining,ifrequired.Experiencedpractitionerscanalsohelpprojectsnavigatepastdeadendsandsteerprojectteamstowardbestpractices.

Alsomore needs to be donewith regard to integrating LinkedData vocabularies in tools for datarecording in the field and laboratory. Like other researchers archaeologists typically show littleenthusiasm to adopt unfamiliar standards and terminology, which is perceived as difficult, time-consuming,andmaynotofferimmediatepracticalbenefits.

Proposed tools thereforeneed to fit intonormalpracticesandhide the semanticapparatus in thebackground, while supporting interoperability when the data is being published. Noteworthy



examplesaretheFAIMSmobiledatarecordingtoolsandtheRightFieldtoolforsemanticannotationoflaboratoryspreadsheetdata.

Recommendations

o Focusonapproachesthatallownon-ITexpertsdomostof theworkofLinkedDatageneration,publicationandinterlinkingwithlittletrainingeffortandexpertsupport.

o Provide useful data mapping recipes and templates, proven tools and guidance material toenablereducingsomeofthetrainingeffortandexpertsupportwhichisstillnecessary inLinkedDataprojects.

o SteerprojectstowardsLinkedDatabestpracticesandprovideadviceonwhichmethodsandtoolsarereallyprovenandeffectiveforcertaindataandtasks.

o CurrentpracticesareverymuchfocusedonthegenerationofLinkedDataofcontentcollections.More could be done with regard to integrating Linked Data vocabularies in tools for datarecordinginthefieldandlaboratory.

PromoteKnowledgeOrganizationSystemsasLinkedOpenData

Briefsummary

Knowledge Organization Systems (KOSs) such as ontologies, classification systems, thesauri andothers areamong themost valuable resourcesof anydomainof knowledge. In thewebof LinkedDataKOSsprovidetheconceptualandterminologicalbasisforconsistentinterlinkingofdatawithinandacrossfieldsofknowledge,enablinginteroperabilitybetweendispersedandheterogeneousdataresources.

TheRDFfamilyofspecificationsprovides“languages”forLinkedDataKOSs.Therelativelylightweightlanguage Simple Knowledge Organization System (SKOS) can be used to transform a thesaurus,taxonomy or classification system to Linked Data. KOSs that are complex conceptual referencemodels (orontologies)ofadomainofknowledgeare typicallyexpressed inRDFSchema (RDFS)ortheWebOntology Language (OWL). LinkedData KOSs aremachine-readablewhich allows variousadvantages. For example a SKOSified thesaurus employed in a search environment can enhancesearch & browse functionality (e.g. facetted search with query expansion), while Linked Dataontologiescanallowautomatedreasoningoversemanticallylinkeddata.

SomeyearsagomanyKOSswerestillmadeavailableascopyrightedmanualsoronlinelookuppages.Recently open licensing of KOSs has become the norm and ever more existing KOSs are beingpreparedandpublishedasLinkedOpenDataforotherstore-use.Followingthepath-breakinglibrarycommunity, the initiative for KOSs as LOD is under way also in the field of cultural heritage andarchaeology. Some international and national KOSs are already available as LOD, Iconclass, Gettythesauri(e.g.Arts&ArchitectureThesaurus),severalUKculturalheritagevocabularies,thePACTOLSthesaurus(France,butmulti-lingual),andothers.

But more still needs to be done for motivating and enabling owners of cultural heritage andarchaeology KOSs to produce LOD versions and align them with relevant others, for examplemapping proprietary vocabulary to major KOSs of the domain. Also more LOD KOSs for researchspecialities,suchastheNomismaontologyfornumismatics,arenecessary.

The sector of cultural heritage and archaeology could also benefit from a dedicated internationalregistryforKOSsalreadyavailableasLODorinpreparation.Anauthoritativeregistrycouldserveas



aninstrumentofqualityassuranceandfosteracommunityofKOSsdeveloperswhoactivelycuratevocabularies.Sucha registrycouldalsoallowannouncingLODKOSsprojects so thatduplicationofworkmaybepreventedandcollaborativeeffortspromoted(e.gvocabularyalignments).

Recommendations

o FostertheavailabilityofexistingKnowledgeOrganizationSystems(KOSs)foropenandeffectiveusage, i.e. openly licensed instead of copyright protected, machine-readable in addition tomanualsandonlinelookuppages.

o Provide practical guidance and suggest effective methods and tools for the generation,publicationandlinkingofKOSsasLinkedOpenData(LOD).

o Encourage institutional owners/curators of major domain KOSs (e.g. at the national level) tomakethemavailableasLOD.

o Promote alignment ofmajor domain KOSs andmapping of proprietary vocabulary, e.g. simpletermlistsortaxonomiesasusedbymanyorganizations,tosuchKOSs.

o Promotearegistry fordomainKOSsthatsupportsqualityassuranceandcollaborationbetweenvocabularydevelopers/curators.

FosterreliableLinkedDataforinterlinking

Briefsummary

The core LinkedDataprinciple arguably is thatpublishers should link their data tootherdatasets,because without such linking there is no “web of data”. In practice this principle is often notfollowed, particularly also not in the field of cultural heritage and archaeology. This means thatalreadyproducedLinkedDataremainsisolated,awebofdatahasnotemergedyet.Thereareseveralreasonsforthisshortcoming.Obviouslyonefactoristhatonlyfewprojectssofarhaveproducedandexposedarchaeological LinkedData.Developersof suchdatawill alsonot considerpopular LinkedData resources like DBpedia/Wikipedia as relevant candidates. Moreover there is the issue ofreliability,thatdataonelinkstowillremainaccessible,whichoftentheyarenot.Surveysfoundthatmanydatasetspresentproblems,forexampleSPARQLendpointsareoftenoff-lineorpresenterrors.

With the increasingnumberof LinkedData resources theirqualityhasbecomeacore topicof thedeveloper community. Detailed quality schemes and metrics are being elaborated and used toscrutinizeresourcesandsuggestimprovements.Thequalitycriteriaessentiallyareabouthowusers(humans andmachines) can discover, understand and access LinkedData resources that arewell-structured,accurate,up-to-dateandreliableover time.Furthermoretheresourcesshouldbewell-documented, e.g. with regard to data provenance and policy/licensing. Ideally the result of thequalityinitiativewillbeeasytousetoolsthatallowLinkedDatacuratorsmonitorresources,detectandfixproblemssothathigh-qualitywebsofdataarebeingdevelopedandmaintained.

The lack of trustworthy resources in many quarters of the “web of data” makes clear that acommunity of curators is necessarywho take care for reliable availability and interlinking of high-quality archaeological LOD datasets and vocabularies. A few domains already have such acommunity, the Libraries and Life Sciences domains, for instance. Also the Ancient World LODcommunityaroundthePelagiosinitiativeortheNomismacommunitycanbementionedasexamplesofgoodpractice.ItappearsthatthedomainofarchaeologyneedsaLODtaskforceandanumberofprojectswhichdemonstrateandmakeclearwhatisrequiredforreliableinterlinkingofLOD.

Recommendations



o Foster a community of LOD curators who take care for proper generation, publication andinterlinkingofarchaeologicaldatasetsandvocabularies.

o FormataskforcewiththegoaltoensurereliableavailabilityandinterlinkingofLODresources;LODqualityassuranceandmonitoringshouldbeestablished.

o Sponsor a number of projects which demonstrate the interlinking and exploitation of someexemplaryarchaeologicaldatasetsasLinkedOpenData.

PromoteLinkedOpenDataforresearch

Briefsummary

LinkedOpenDatabasedapplicationsthatdemonstrateconsiderableadvancesinresearchprocessesand outcomes could be a strong driver for a wider uptake of the LOD approach in the researchcommunity.CurrentexamplesofLinkedDatauseforresearchpurposesrarelygobeyondsemanticsearch and retrieval of information. This has not gone unnoticed by researchers who expectrelevanceofLinkedOpenDataalsoforgeneratingandvalidatingorscrutinizingknowledgeclaims.Toallowforsuchusesatighterintegrationofdiscipline-specificvocabulariesandeffectiveLinkedDatatoolsandservicesforresearchersarerequired.

Expectationsofreseach-focusedapplicationsofLODinthefieldofculturalheritageandarchaeologyoften relate to the CIDOC CRM as an integrating framework. The CIDOC CRM is recognised as acommon and extendable ontology that allows semantic integration of distributed datasets andaddressingresearchquestionsbeyondtheoriginal,localcontextofdatageneration.Notably,intheARIADNE project several extensions of the CIDOC CRM have been created or enhanced, e.g.CRMarchaeo,anextensionforarchaeologicalexcavations,andextensionsforscientificobservationsandargumentation(CRMsciandCRMinf).

Tomeet expectations such as automatic reasoning over a largeweb of archaeological datamanymore(consistent)conceptualmappingsofdatabasestotheCIDOCCRMwouldbenecessary.LinkedData applications then might demonstrate research dividends such as detecting inconsistencies,contradictions, etc. in scientific statements (knowledge claims) or suggesting new, maybeinterdisciplinarylinesofresearchbasedonsurprisingrelationshipsbetweendata.

Recommendations

o LODbasedapplicationsthatenableadvancesinarchaeologicalresearchprocessesandoutcomesmayfosteruptakeoftheLODapproachbytheresearchcommunity.

o LODbasedapplicationsforresearchwillhavetodemonstrateadvantagesoverorotherbenefitsthanalreadyestablishedformsofdataintegrationandexploitation.

o DevelopLODbasedservicesthatgobeyondsemanticsearchandretrievalofinformationandalsosupportotherresearchpurposes.

o BuildontheCIDOCCRMandavailableextensionstoexploitconceptuallyintegratedLOD.



2.2.5 LinkedDatadevelopmentinARIADNE

Briefsummary

The developmental ARIADNE Linked Data work described in this chapter has focused on theproduction of (and support for) SKOS subject vocabularies,mappings between those vocabulariesandtheArt&ArchitectureThesaurus,inordertoprovideamultilingualcapability,andthemappingsof datasets to the CIDOC-CRM. Furthermore three advanced case studieswith demonstrators arepresentedthatgenerateanduseLinkedDatabasedontheCIDOCCRMandkeysubjectvocabularyhubs:coins,woodenmaterialandsculptures.

The first two case studies involve information extraction from text reports in addition tomappingdatasets, while the third explores external linking beyond the immediate ARIADNE datasets.ExploratoryworkonminingofLinkedDataandNLPtechniquesaredescribedbutbothareresearchareas with potential for much further work. The transformation of the metadata of the datasetsregisteredintheARIADNEdatacataloguetoLinkedDataisdescribedinthenextchapter,asarethedetailsoftheARIADNELinkedDataservice.

The demonstrators are still being finalised at the time of this deliverable butwill be available forgeneraluseviatheARIADNEPortal.Forthereasonsdiscussedintheearlychapters,thecasestudiesareexperimentalinvestigationsofthefutureusecasesthatareaffordedbyLinkedDatatechnology;they result in (working) research demonstrators rather than actual operational systems. Theyillustrate thekindsofpossibilities forcrosssearchandthesemantic integrationofdiversekindsofdatasetsandtextreportsthatLinkedDataandtherelatedsemantictechnologiesmakepossible.

Oneobviousfindingfromtheexperiencetodateisthecriticalimportanceofthesubjectvocabularies(e.g.theAAT)combinedwiththeCIDOCCRMontologyentities,whichactaslinkinghubsinthewebofdata.MoreworkisneededontheidentificationoffurtherlinkinghubsandconsequentsemanticenrichmentoftheLinkedDatatorelevantexternaldatasets.Oneexampleofapotentiallinkinghubis thePeriod0setofculturalperiodswhichcanbeusedbyprovidersofvariousarchaeologicalandotherculturalheritagedatasets.

Necessary for thewidespreaduptakeof theLinkedDataapproach is theavailabilityofavarietyofmapping and alignment software for different contexts, together with evaluative studies andguidelinesastotheiruse.Beyondthat,tomotivateuserorganisationstodevotescarceresourcestoworkingwithLinkedData,someexemplarworkingapplicationsareneededthataddressarealuser(scientific/research)need.Suchapplicationsshouldofferauserinterfacethatiseasyandattractivetoworkwith,onethatdoesnotrequireprogrammingskillsordetailedknowledgeoftheunderlyingdataschemaorontologystructure.

Itshouldnotnecessarilybeassumedthattheend-applicationdirectlyoperatesovera(LinkedData)triplestore.Thereareadvantagesindoingsofordataupdatesandexternalconnectionsanditisanobviousroute.However,periodicharvestingofLinkedDataisapossibilityforapplicationsthathavereasons toemployawider rangeofprogrammingplatforms.Anotherpossibility is for LinkedDataproviderstoconsiderexposingprogrammaticwebservicesforapplicationdevelopers(inadditiontoa SPARQL endpoint), assuming that an appropriate set of of use cases for the services can beidentified.

Lessonslearned

o MappingofdatasetstoestablisheddomainKOSs(inourcaseCIDOCCRM,AATandothers)allowstheirintegrationwithinandbeyondthecatalogueofadataportal.



o State-of-the-art linking hubs will play an increasingly important role in the web of LOD,comprehensivedomainthesauriastheAATaswellasspecialisedvocabulariesliketheNomismathesaurus.

o The mapping of datasets to such hubs requires domain knowledge, easy to use tools, andguidance of users who carry out such work for the first time. While recommender tools arehelpful,fullyautomatedmappingappearsunlikelytoachivequalityresultsatthecurrenttime.

o TheARIADNEportalandpilotdemonstratorsshowthatthisworkisworththeeffort.Butthereisstill away to go before advanced uses of LODwill become applicable and beneficial in onlineresearchenvironments;moreeffortmustbeinvestedtomakethishappen.

o ThereismuchscopetoexploretheutilityofLODinpractice,takingaccountoftheobjectivesandrequirementsofdifferentusercommunities.ThebestwaystoprovideandemployLODwilllargelydependon their specific contexts (museumcollections, dataarchivesor researchplatforms, forinstance), together with the anticipated use cases. In order to motivate user organisations towork with Linked Data, exemplar working applications that address a real user(scientific/research)needwouldbeveryhelpful.

2.2.6 ARIADNELODCloud

Briefsummary

TheARIADNEregistryholdsmetadataofdataresourcesfromthecontentproviders.Thesemetadataare being collected and enriched with an aggregator (MORe) and included in the ARIADNE datacatalogue. ARIADNEmakes the catalogue and other data generated in demonstrators available asLinkedOpen Data (LOD); thereby the ARIADNE LOD can become part of aweb of Linked Data ofarchaeologicalandrelatedotherinformationresources.

ThisworkwithinARIADNEinvolvedtheuseofasuitableRDFstoreandgraphdatabasefortheLinkedData generation and linking efforts. The project has experimented with two such technologies,Virtuoso and Blazegraph, to perform archaeologically relevant SPARQL queries on the generatedLinked Data, and to allow updates of datasets using the SPARQL 1.1 Graph Store HTTP Protocol.Based on this preliminary work, a scalable implementation that can efficiently support thepublication and use of the ARIADNE LOD has been designed and realized to offer three differentservices:theLinkedOpenDataServer,theDemonstrators,andtheMappingandOntologyServer.

The Linked Open Data Server provides access to a large RDF dataset, which comprises of severalgraphs of archaeological datasets and can be queried via a SPARQL endpoint. The DemonstratorshavebeendevelopedtoexemplifythecapabilityofLinkedDatabaseditem-leveldataintegrationtosupportansweringarchaeologicalresearchquestions.Theyrepresentthreedifferentsubjectareasofarchaeology: coins, sculptures and wooden material. For each a number of datasets have beenintegratedbasedonmappingstotheCIDOCCRM(andrecentextensions)anduseofotherdomainvocabularies.TheMappingandOntologyServerprovides informationabout themappingsand thevocabularies(ontologies,thesauri)involvedintheARIADNELODCloud.

ThecurrentARIADNELODCloudis justtheinitialstageofan informationspacethat isexpectedtogrow in terms of data, vocabularies, services and users. Experiments to exploit the ARIADNE LODhave juststarted,withpromisingresultsasshownby theDemonstrators.Planned futureworkwillaim to proceed with linking the available Linked Data to relevant other datasets. To promoteinterlinking, theARIADNE LODwill be announced via relevantmailing lists, newsletters etc. of the



LinkedDatacommunity in the fieldofarchaeologyandculturalheritage.AnumberofLinkedDatadevelopers will also be contacted directly to suggest and discuss interlinking with their or otheravailabledatasetsinthewebofLOD.

Lessonslearned

WhiletheLinkedOpenDatastandardsareessentialforintegratingdata,thetechnologysupportingsuchintegrationisstillinitsinfancy.TheARIADNELOD,comprisingofLODoftheARIADNEcatalogue,three demonstrators and various vocabularies sum up to about 32million RDF triples.While anyrelational database can easily handle millions of records, the corresponding amount of RDF in acurrenttriplestorecancauseseriousefficiencyproblemsasexperiencedintheexperimentationwiththe ARIADNE LinkedData Cloud. It is becoming apparent that this is the price to be paid to haveinteroperability. More robust and efficient graph databases are required if we want to proceedtowardsBigDataasLinkedData.ThisisthefirstlessonthatwehavelearnedwhileimplementingtheARIADNELinkedDataCloud.

Thesecondlessoncomesfromthegraphdatamodel.Thismodelisintrinsicallybinary,hencemakesitdifficulttoexpresshigherrankrelations,andtoeasilyimplementdataconnectionpatterns.Inthelatter case, the patterns may involve data chains that span several arcs, and their definition andimplementationisnottrivial.Conversely,correlationsbetweendataitemscanbeepitomizedbysuchpaths,whichneedtobedetected,andthis isacomputationallyvery intensivetask if the lengthofthepathsgobeyond2-3arcs.Thisfacthasalwaysbeenknownfromatheoreticalpointofview,butworkingwithrealdatawecouldexperienceitinpractice.



3 LinkedOpenData:Backgroundandprinciples

This chapter introduces the Linked Open Data approach, describing the development of theapproach, the LinkedData principles, standards and good practices for datasets and vocabularies.The chapter also suggests what adopters of the Linked Data approach should consider first, anddescribesthemainstepsintheLinkedDatalifecycle.

3.1 LOD–Abriefintroduction

LinkedDataareWeb-baseddata thataremachine-readableand semantically interlinkedbasedonWorldWideWeb Consortium (W3C) recommended standards, in primis the Resource DescriptionFramework(RDF)familyofspecificationsbutalsoothers.LinkedOpenDataaresuchdataresourcesthatarefreelyavailableunderanopenlicense(e.g.CreativeCommonsAttribution-CC-BY)orinthePublicDomain.

The LinkedData standards allow the creation, publication and linkingofmetadata and knowledgeorganizationsystems(KOSs)inwaysthatmakethesemantics(meaning)ofdataelementsandtermscleartohumansandmachines.LinkedDataarelinkedsemanticallybasedonexplicit,typedrelationsbetweenthedataresources.

The semanticwebof LinkedDataessentially is about relationshipsbetween information resourcessuchascollectionsofdigitalcontent.Themetadataofdigitalcollections(orothersetsofdataitems),describe different facets of the resources, e.g. what, where, when, who, etc. For such facetsknowledgeorganizationsystems(KOSs)suchasthesauriprovideconceptsandterms.

TheW3C recommended LinkedData standardsprovide thebasisof a semanticweb infrastructurethat facilitates domain-independent interoperability of data. Building on the standards, domain-basedmetadataandknowledgemodelsareneeded toenable interoperabilityand rich interlinkingbetweendataofspecificdomainssuchasculturalheritageandarchaeologicalresearch.

The requirements for semantic interoperability are considerable. In the case of data sets ofarchaeologicalprojects,storedindifferentdigitalarchives,themetadataofthedatapackagesmustbe converted to Resource Description Framework (RDF) and include terms of shared vocabulary,which alsomust be available as LinkedData (e.g. in the Simple KnowledgeOrganization System–SKOSformat).Datacuratorsthusneedtobecomefamiliarwithnewstandardsandtoolstogenerate,publish and connect Linked Data. But it does no mean that they must abandon establisheddatabases,becausetoolsareavailabletooutputRDFdatafromexistingdatabases(RDB2RDFtools).

Buildingsemantice-infrastructureandservicesforaspecificdomainrequirescooperationbetweendomain data producers/curators, aggregators and service providers. Cooperation is necessary notonly for sharing datasets through a domain portal (i.e. the ARIADNE data portal), but also to usecommonoralignedvocabularies(e.g.ontologies,thesauri)fordescribingthedatasothatitbecomesinteroperable.Forexample,inARIADNEthedataprovidersagreedtomapvocabularywhichtheyusefor their dataset metadata to the comprehensive and multi-lingual Art & Architecture Thesaurus(AAT),whichisavailableasLinkedOpenData.

ARIADNEalsorecommendstheCIDOCConceptualReferenceModel(CRM)asacommonontologyfordata integration based on Linked Data. The CIDOC CRM has been developed specifically fordescribing cultural heritage knowledge and data. Archaeology partly overlapswith this domain aswellasneedsmodellingofadditionalconceptualknowledge,forexample,todescribeobservationsofanexcavation(e.g.stratigraphy).TheARIADNEReferenceModelcomprisesthecoreCIDOCCRM



and a set of enhanced and new extensions, including for the archaeological excavation process(CRMarchaeo)andbuiltstructuressuchashistoricbuildings(CRMba)1.

3.2 Historicalandcurrentbackground

Thebasic conceptof LinkedDatahasbeendefinedbyTimBerners-Lee, the inventorof theWorldWideWeb, in anarticlepublished in2006 (Berners-Lee2006). The concepthelped to re-orientateandchanneltheinitialgrandvisionoftheSemanticWebintoaproductivenewavenue.Inanupdate2010oftheinitialarticleonLinkedOpenDataBerners-LeealigneditwiththeOpenDatamovement(Berners-Lee2010).

In a historical perspective it is worth noting that Berners-Lee since 1998 had addressed various“Design Issues” of the SemanticWeb on thewebsite of theWorldWideWeb Consortium –W3C(Berners-Lee1998-). In2001thevisionofaSemanticWebreachedawideraudiencewithahighlyinfluentialarticleintheScientificAmerican(Berners-Lee,Hendler&Lassila2001).Thewidelyquoted“SemanticWebStatement”ofthededicatedW3CActivity(startedin2001)included:“TheSemanticWebisavision:theideaofhavingdataonthewebdefinedandlinkedinawaythatitcanbeusedbymachines not just for display purposes, but for automation, integration and reuse of data acrossvariousapplications”.2

Previous to Berners-Lee’s Linked Data article (2006) the research and development communitypresented the SemanticWeb vision as a complex stack of standards and technologies. This stackseemedalways“underconstruction”andtogetherwith thedifficult tocomprehendSemanticWebterminologycreatedtheimpressionofanacademicactivitywithlittlerealworldimpact.

There-brandingoftheSemanticWebasLinkedDataandthemoderatedefinitionofsuchdatawasabrilliantcommunicativecoup.Itsignalledare-orientationwhichwaswelcomedbymanyobservers,includingbusiness-orientedinformationtechnologyconsultants(e.g.PricewaterhouseCoopers2009;Hyland 2010). In 2009, a paper co-authored by Berners-Lee on “Linked Data – the story so far”summarised: “The term LinkedData refers toa set of best practices for publishingand connectingstructureddataontheWeb.Thesebestpracticeshavebeenadoptedbyanincreasingnumberofdataprovidersoverthelastthreeyears,leadingtothecreationofaglobaldataspacecontainingbillionsofassertions-theWebofData”(Bizer,Heath&Berners-Lee2009).Howevertheauthorsalsonotedsome issues inLinkedData, inparticular, thequalityandopen licensingofLinkedDatarequiredtoallowfordataintegration.

In 2010 Berners-Lee’s request for Linked Open Data aligned the Linked Data with the Open Datamovement (Berners-Lee2010),whichhasbecomeparticularly strong in the governmental / publicsector.InthissectorOpenDataareseenasameanstoensuretrustthroughtransparencyandmakepublicly funded informationavailable (Huijboom&VandenBroek2011;Geiger&Lucke2012)3. Inthis context Linked Open Data are recognized as just the right approach to expose and connect

1DescriptionoftheARIADNEReferenceModelandindividualextensions(includingreferencedocument,

presentation,RDFSencoding)isavailableathttp://www.ariadne-infrastructure.eu/Resources/Ariadne-Reference-Model

2SinceDecember2013,theW3CSemanticWebActivityissubsumedundertheW3CDataActivitywhich“hasalargerscope;neworcurrentWorkingandInterestGroupsrelatedto‘traditional’SemanticWebtechnologiesarenowpartofthatActivity”(http://www.w3.org/2001/sw/).Inthecourseofthisshift,thequoted“vision”statementhasbeenremoved(replacedbysomeother,rathervaguelines).

3TheinternationaldevelopmentofopengovernmentaldataistrackedandmeasuredbytheOpenDataBarometerproject,http://opendatabarometer.org



existinglegacydatasilosaswellasenablere-useofdatafornewservices.Thesamerationaleappliestotheculturalheritagesectorwithitsheavilypublicly-fundedinstitutions.

The Open Datamovement has also renewed and strengthened the interest of governmental andpublicsectorinstitutionstoimproveandintegratetheirknowledgeorganizationsystems(KOSs).Onemajor goal here is enabling access to governmental, cultural and scientific information resourcesacrossdifferentorganizationaldepartments,institutionsanddomains(Hodge2014).

3.3 LinkedDataprinciplesandstandards

3.3.1 LinkedDatabasics

In 2006, Berners-Lee published the basic article on Linked Data in which he summarised in fourprinciples how to “grow” the Semantic Web (Berners-Lee 2006). In these principles UniformResourceIdentifiers(URIs)andtheW3CResourceDescriptionFramework(RDF),whichrequirestheuseofURIs,arekeystandardstofollow,whichwedescribeinacommentarytoBerners-Lee’sLinkedDataprinciplesbelow.Thebasicprinciplesare:

1. UseURIsasnamesforthings.

2. UseHTTPURIssothatpeoplecanlookupthosenames.

3. When someone looks up a URI, provide useful information, using the standards (RDF,SPARQL).

4. IncludelinkstootherURIs,sothattheycandiscovermorethings.

Thissoundssimple,butwhataretheseURIs,RDFandSPARQL?

URIs: Linked Data use Uniform Resource Identifiers4 as globally unique identifiers for any kind oflinkable “resources” such as abstract concepts or information about real-world objects. Moreprecisely,LinkedDatashouldusedereferencableHTTPURIs,whichallowawebclientlookupanURIusing the HTTP protocol and retrieve the information resource (content,metadata, description ofterm,etc.).URIsarethekeyelementofLinkedDatastatementswhichareformedaccordingtotheRDFmodel(seebelow).It is importanttodesignandserveURIsproperly,followingbestpractices.5The persistence of URIs is a crucial part of the whole setup of the “web of data”, especiallyconcerningtherequiredtrustinthereliabilityofLinkedDatasources.

RDF: Linked Data is based on the W3C Resource Description Framework (RDF) model.6 The RDFmodel uses subject-predicate-object statements (the so called “triples”) which employ derefer-encableURIs fordescribingdata items.ThepredicateofanRDFstatementdefines thepropertyofthe relation that holds between two items. This allows for setting typed links between the itemswhichmakeexplicitthesemanticsoftherelations.AsearchablewebofLinkedDatacanbecreatedifdata providers publish the items of their datasets as HTTP URIs and related items are connected

4UniformResourceIdentifier(URI):GenericSyntax,RFC3986/STD66(2005)specification,

http://tools.ietf.org/html/std66;W3C(2004)Recommendation:ArchitectureoftheWorldWideWeb(Volume1),15December2004,http://www.w3.org/TR/webarch/#identification

5W3C(2008):CoolURIsfortheSemanticWeb,http://www.w3.org/TR/cooluris/;the“10rulesforpersistentURIs”suggestedinISA(2012);andArwe(2011)onhowtocopewithun-coolURIs.

6W3C(2014)Recommendation:RDF1.1ConceptsandAbstractSyntax,25February2014,https://www.w3.org/TR/rdf11-concepts/



through links of RDF statements. For example, one dataset may contain information aboutarchaeologicalsites inaregion,anotherdatasetaboutdatadepositsofexcavations,anotheraboutarchaeologistssothatonecansearchatwhichsitesexcavationshavebeenconducted,wherewhatkindofthedataisavailable,whofrominstitutionswasinvolved,etc.

SPARQL: The SPARQL Protocol and RDF Query Language (SPARQL)7 allows for querying andmanipulatingRDFgraphcontent inanRDFstoreorontheWeb, includingfederatedqueriesacrossdifferentRDFdatasets.

3.3.2 LinkedOpenData

In2010,Berners-Leeaddedasectionon“IsyourLinkedOpenData5Star?”totheLinkedDataarticleof2006(Berners-Lee2006).Thissectionaddressedthemissingprincipleofopennessofthedata.

Berners-Lee’s5starschemeofLinkedOpenData8:

* Availableontheweb(whateverformat)butwithanopenlicence,tobeOpenData

** Availableasmachine-readablestructureddata(e.g.excelinsteadofimagescanofatable)

*** as(2)plusnon-proprietaryformat(e.g.CSVinsteadofexcel)

**** Alltheaboveplus,UseopenstandardsfromW3C(RDFandSPARQL)toidentifythings,sothatpeoplecanpointatyourstuff

***** Alltheabove,plus:Linkyourdatatootherpeople’sdatatoprovidecontext

Somecommentsmaybeappropriatetorelatethisschemetothe2006definitionofLinkedDataandexplainsomepointswhichmaybemisunderstood:

Available on theweb (whatever format): Thephrase “on theweb” as used in the SemanticWebcommunity does not necessarilymean a webpage, but any information resource that has an URI(UniformResourceIdentifier)andcanbelinkedandaccessedand,possibly,actedupon.Howeverthestandard example is a simple HTML page that presents information and includes links to othercontent(e.g.storedonalocalserver).(whateverformat):Meansthatatthefirst,1-starlevelorsteptowardsLinkedOpenDataitisnotseenasimportantthatthecontentmaybedifficulttore-use(e.g.aPDFofatextdocumentoraJPEGimageofadiagram).

Open licensing: Concerning the important issue of explicit open licensing Berners-Lee notes: “Youcanhave5-starLinkedDatawithoutitbeingopen.However,ifitclaimstobeLinkedOpenDatathenitdoeshavetobeopen,togetanystaratall.”Hedoesnotsuggestanyparticular“openlicense”likeCreativeCommons(CC0,CC-BYandothers)9orOpenDataCommons(PDDL,ODC-By,ODbL)10.

7W3C(2013)Recommendation:SPARQL1.1Overview,21March2013,http://www.w3.org/TR/2013/REC-

sparql11-overview-20130321/8Seealsothe“5★OpenData”websitewhichprovidesmoredetailandexamples,http://5stardata.info9CreativeCommons,https://creativecommons.org/licenses/10OpenDataCommons,http://opendatacommons.org/licenses/



Machine-readable structured data: In contrast to the first statement “(whatever format)”, hereBerners-Leeemphasisesthatthedatashouldnotbe“canned”(i.e.notanimagescan/PDFofatable)butopenforre-usebyothers(i.e.theactualtableinExcelorCSVdata).

Non-proprietaryformat:Thiscriterionisaboutpreventingdependenceonproprietarydataformatsand software to read the data. However it is somewhat at odds with the widespread use ofproprietaryformatssuchasExcelspreadsheets.Forexample,manypotentialuserswillbecapableofre-using such spreadsheets, and it is unlikely that data providerswould convert their data to CSV(CommaSeparatedValues)justtocomplywiththecriterion.Thereforetheprimarycriterionisthatthedatashouldnotbe“canned”and,secondary,providedinaneasytore-useformat.

Useopen standards fromW3C (RDFand SPARQL) to identify things, so that people canpoint atyour stuff:While the criteria above address theopennessof data/content in termsof format andlicense,hereweentertherealmofLinkedData,e.g.URIs“toidentifythings,sothatpeoplecanpointatyourstuff”whentheyformRDFstatements(asdescribedinthesectionabove).

Link your data to other people’s data to provide context: The highest level of LinkedOpenDatademandsinterlinkingthroughRDFowndatawithotherLinkedDataresourcestocreateanenrichedweb of information. The RDF links connect data from different sources into a graph that enablesapplications (e.g. a LinkedDatabrowser) to navigate between themanduse their information forprovidingservices.

Insummary:

• Thecriteria forearning the first three stars relate to “opendata” in termsofdata formatandlicensing; notably the first three stars can be earned without employing W3C standards andtechniques.

• Thenext level, 4-stardata clearlypoints to these standardsand techniques (RDF, SPARQLandothers),while5-stardatarequires interlinkingowndatawithresourcesofotherssothatarichwebofdatacanemerge.

• Surprisingly, Berners-Lee did not address metadata and knowledge organization systems,althoughtheycanbesubsumedunder“structureddata”.However,inresponsetosomecriticismheadded:“Yes,thereshouldbemetadataaboutyourdataset.Thatmaybethesubjectofanewnoteinthisseries.”

• Toemphasiseagaintheimportanceofopenlicensing,Berners-Leestates:“LinkedDatadoesnotofcourseingeneralhavetobeopen(…).Youcanhave5-starLinkedDatawithoutitbeingopen.However,ifitclaimstobeLinkedOpenDatathenitdoeshavetobeopen,togetanystaratall.”

3.3.3 MetadataandvocabularyasLinkedData

Above we noted that Berners-Lee’s Linked Open Data principles do not mention metadata andknowledge organization systems (KOSs), arguably to avoid addressing such more formalizedstructures of Linked Data. They come in two variants of “vocabularies”: 1) metadata schema forcontent collections, and 2) knowledge organization systems (KOSs) that provide concepts formetadatarecordsofcollectionitems.

Metadataschemasdefineasetofelements(andproperties)fordescribingtheitems.Forexample,the 15 elements of the Dublin Core Metadata Element Set (e.g. creator, title, subject, publisher,etc.)11areoftenusedformetadatarecordsofculturalproducts.KOSs(e.g.thesauri)arebeingused11DublinCoreMetadataElementSet,Version1.1,2012-06-14,http://dublincore.org/documents/dces/



to select values for the element fields in metadata records (e.g. the subject/s of a paper). ThestructureandcontentofbothmetadataschemasandKOSscanberepresentedasLinkedData.

AmongtheKOSs,thesauriandclassificationssystems(ortaxonomies)aremostlyrepresentedintheW3CSimpleKnowledgeOrganizationSystem(SKOS)format12.Athesaurusinthisformatcanbeusedto state thatoneconcepthasabroaderornarrowermeaning thananother,or that it is a relatedconcept,orthatvarioustermsarelabelsforagivenconcept.

KOSs thatarecomplexconceptual referencemodels (orontologies)ofadomainofknowledgearetypicallyexpressedinRDFSchema(RDFS)13ortheWebOntologyLanguage(OWL)14,whichallowforsomeautomatedreasoningoverthesemanticallyinterlinkedresources.

Besides thementionedKOSs, therearegazetteersofgeographical locations (e.g.GeoNames15)andsocalledauthorityfilesofmajorinstitutions,forexample,fornamesofpersons(e.g.VIAF)16.Atthelowestlevelofcomplexityareflatlistsoftermsandglossaries(termlistsincludingdescriptionoftheterms).

3.3.4 GoodpracticesforLinkedDatavocabularies

Because of the core role of knowledge organization systems (KOSs) for Linked Data, developersrecommendadditionalgoodpractices for suchvocabularies (e.g.Heath&Bizer2011 [section5.5];W3C 2014 [vocabulary checklist]). Vocabularies should of course follow the basic Linked Dataprinciples, e.g. use dereferenceable HTTP URIs so that clients can retrieve descriptions of theconcepts/terms17. The first specific rule for vocabularies is to re-use or extend wherever possibleestablishedvocabularybeforecreatinganewone.Therationaleforre-useisthatdifferentresourcesonthewebofLinkedDatawhicharedescribedwiththesamevocabularytermsbecomeinterlinked.Thismakesiteasierforapplicationstoidentify,processandintegrateLinkedData.

Moreover, re-use and extension of existing vocabularies can lower vocabulary development costs.Extensionheremeansthatvocabularydevelopersre-usetermsfromoneormorewidelyemployedvocabularies (which usually represent common types of entities) and define proprietary terms (intheirown“namespace”)forrepresentingaspectsthatarenotcoveredbythesevocabularies.

It is generally recommended that publishers of Linked Data sets (e.g. metadata of contentcollections),shouldalsomaketheiroftenproprietaryvocabulary(e.g.thesaurus,termlist)availableinLinkedData format.As Janowiczetal. (2014)note,“queryingLinkedDatathatdonot refer toavocabulary is difficult and understanding whether the results reflect the intended query is almostimpossible”.Theauthorssuggesta5-starratingforvocabularies:

o One star is assigned if a Web-accessible human-readable description of the vocabulary isavailable(e.g.awebpageorPDFdocumentingthevocabulary),

12W3C(2009)Recommendation:SKOSSimpleKnowledgeOrganizationSystem,18August2009,

https://www.w3.org/2004/02/skos/13W3C(2014)Recommendation:RDFSchema1.1,25February2014,http://www.w3.org/TR/rdf-schema/14W3C(2012)Recommendation:OWL2WebOntologyLanguageDocumentOverview(SecondEdition),11

December2012,https://www.w3.org/TR/2012/REC-owl2-overview-20121211/15GeoNames,http://www.geonames.org16VIAF-VirtualInternationalAuthorityFile(combinesmultiplenameauthorityfilesintoasinglename

authorityservice),https://viaf.org17W3C(2008)WorkingGroupNote:BestPracticeRecipesforPublishingRDFVocabularies,28August2008,

https://www.w3.org/TR/swbp-vocab-pub/



o Two stars can be earned if the vocabulary is available in an appropriate machine-readableformat,forinstanceathesaurusinSKOSformatoranontologyinRDFSorOWL,

o Three stars will receive a vocabulary that also has links to other vocabularies (for example, amappingbetweenproprietarytermstocorrespondingtermsofwidelyemployedthesauri),

o Four stars are due if also machine-readable metadata about the vocabulary is available (e.g.author/s,vocabularylanguage,version,license),

o Finally, 5 stars are reserved if the vocabulary is also linked to by other vocabularies, whichdemonstratesexternalusageandperceivedusefulness.

The criteria for the third and fifth star concern linking of vocabularies. Such linking requires thatvocabulary owners/publishers produce a mapping between their vocabulary concepts/terms,ontologyclassesorpropertiesandothervocabularies,whichshouldbedonebysubjectexperts. InthecaseofthesauriinSKOSformatsuchmappingsforexampleareskos:exactMatch(twoconceptshave equivalent meaning), skos:closeMatch (similar meaning), skos:broadMatch andskos:narrowMatch(broaderornarrowermeaning).ForontologiesRDFSchema(RDFS)andtheWebOntologyLanguage(OWL)definelinktypeswhichrepresentcorrespondencesbetweenentityclassesandproperties(e.g.rdfs:subClassOf,rdfs:subPropertyOf).

3.3.5 MetadataforsetsofLinkedData

Linked Data resources are assets which, like any other valuable information resource, should bedescribed with machine-processible metadata. Linked Data resources include data, metadata andvocabularies,andlinksestablishedbetweenthem(link-sets).Forexample,amappingbetweentwovocabularies isavaluable link-setwhichshouldbedocumentedwithmetadataandprovidedtoanappropriate registry. Themetadata shouldprovidedescriptive, technical,provenanceand licensinginformationsuchas:

o What kind of resource is available in terms of content, format, etc. (e.g. a thesaurus, in SKOSformat,serializedinJSON18),

o Whocreated/providesit(author/s,publisher)andotherprovenanceinformation(e.g.version,lastupdateetc.),

o Licensing:explicitlicenseorwaiverstatementsshouldbegiven;forLOD“openlicenses”suchasCreativeCommons (CC0,CC-BY)orOpenDataCommons (PDDL,ODC-By)canbeconsideredasadequate,

o Where and how can the resource be accessed (e.g. an HTML webpage, RDF dump, SPARQLendpointforqueryingthedata).

Onewidely used vocabulary for describing RDF datasets and links between them (link-sets) is theVocabularyofInterlinkedDatasets-VoiD(Alexanderetal.2009)19.Schmachtenbergetal.(2014a)intheir survey of the Linked Open Data Cloud in 2014 found that of 1014 identified datasets 140(13.46%) were described with VoiD. Most users of VoID were providers of Linked Data in thecategoriesGovernment,Geographic, and Life Sciences. In thehumanities forexample thePelagiosinitiative for linking of Ancient World resources based on the places they refer requests data

18JSON-JavaScriptObjectNotation(isalightweightdata-interchangeformat),

https://en.wikipedia.org/wiki/JSON19W3C(2011)InterestGroupNote:DescribingLinkedDatasetswiththeVoIDVocabulary,3March2011,

http://www.w3.org/TR/void/



providerstomakeavailableaVoIDfile;thefiledescribesthedataset(mappingsofplacereferencestooneormoregazetteers),publisher,licenseetc.,andcontainsthelinkfromwhichPelagioscangetthedateset20.

TheNetworkedKnowledgeOrganization Systems (NKOS) TaskGroupof theDublinCoreMetadataInitiative(DMCI)hasbeenworkingonaDublinCorebasedmetadataschemaforvocabularies/KOSs.OneimportantfunctionofthisschemaisdescriptionofKOSsinvocabularyregistriesorrepositories(Golubetal.2014).ThesuggestedDublinCoreApplicationProfile-NKOSAPhasbeenreleasedfordiscussion in2015(Zeng&Žumer2015).ForprovidingmetadataofontologiestheVocabularyofaFriend(VOAF)21 isoftenbeingused.Forexample,theLinkedOpenVocabularies(LOV)registryusesVOAF (and dcterms) for describing registered ontologies, i.e. vocabularies in RDFS or OWL(Vandenbusscheetal.2015).

3.4 Whatadoptersshouldconsiderfirst

Adopters of the Linked Data approach should first think about what they wish to achieve bypublishingoneormoredatasetsasLinkedData.IfthegoalisprimarilymakingdataavailableasOpenData thereare simpler solutions, forexampleproviding thedataasadownloadableCSV file22. ForLinkedDatathegoalgenerallyisenrichmentofdataandservicesbyinterlinkingowndatawithdataofotherproviders.Adoptersthereforeshouldconsiderwhichowndatawillgeneratemostvalue ifavailableasandinterlinkedwithotherLinkedData.

Linked Data should not be published “just in case”. Rather publishers should consider the re-usepotential and intended or possible users of their data. As Linked Data consumers they need toaddressthequestionofwhichdataofotherstheycouldlinkto.

Thesequestionsmakecleartheimportanceofjointinitiativesforprovidingandinterlinkingdatasetsofcertaindomains.Particularlysmallinstitutionsshouldlookforandconnecttoarelevantinitiative.A framework for collaboration on LinkedData can ensure value generation, for example, by usingcommon vocabularies. Linked Data developers should also ensure institutional commitment andsupport, i.e. an official project with a clear mandate, allocated staff and resources (cf. Smith-Yoshimura2014f).

LinkedDataadoptersofallsizeswillbeststartwithasmalltargetedprojectthatdoesnotrequirealotof resources.Theprojectshouldallowgaining first-handexperience inLinkedDataandprovidepotentialfortakingnextsteps.ObviouslycreatingHTTPURIsfortheselecteddataisanessentialsteptowardsinterlinkingitbasedonRDF.ExposinglocaldataidentifiersasHTTPURIsallowsopeningupadatabasesothatotherscanlinktoandreference/citethedata.

LargeinstitutionssuchasgovernmentalagenciesmaybenefitfromstreamliningwiththeLinkedDataapproachinternalprocessesforsharingandintegrationofdataofdifferentdepartmentsandcloselyrelated organisations. Such institutions are also often those which publish major controlledvocabularieswhichotherscanusetoconnectdata(Archeretal.2014:55-56).

20Pelagios:JoiningPelagios,https://github.com/pelagios/pelagios-cookbook/wiki/Joining-Pelagios21VOAF-VocabularyofaFriend,http://lov.okfn.org/vocommons/voaf/v2.3/22SeeHeath(2010)foracomparisonbetweenprovidingaCSVfilevs.LinkedData.



3.5 MasteringtheLinkedDatalifecycle

Theprevioussectionspresenttheprinciples,standardsandgoodpracticesofLinkedData,butdonotdescribehowsuchdataareactuallygenerated,publishedandinterlinked.Thisstudydoesnotintendprovidingaguidebookformasteringthesocalled“lifecycle”ofLinkedData,thedifferentstepsthatarenecessarytogettoandbenefitfromsuchdata.Inbrief,themainstepsare:

o Selectarelevantdataset:ChoseadatasetwhichallowsgeneratingvalueifmadeavailableasRDFdataandlinkedtootherLOD,includinglinkingofthedatasetbyothers.Thepublishershouldofcoursebeabletoprovidethedataunderanopenlicenseorplaceitinthepublicdomain.

o Cleanandpreparethesourcedata:Bringthesourcedatainashapethatitiseasytomanipulateand convert to RDF, addressing issues of data quality such as missing values, invalid values,duplicaterecords,etc.TheOpenRefine23toolisrecommendedforthistask.

o DesigntheURIsofthedataitems:FollowsuggestedgoodpracticefordesigningthestructureoftheURIs(e.g.W3C2008;ISA2012).

o Define the target datamodel:Re-use an existingmodel that is beingused in thedomain (e.g.CIDOCCRM for cultural heritage data) or create one re-using concepts fromwidely employedvocabularies;re-usewillaiddatainteroperabilityanddecreasedevelopmenteffort/costs.

o TransformthedatatoRDF:Inthetransformationthesourcedata(e.g.datatables)areconvertedtoasetofRDFstatements(graph-basedrepresentation)accordingtothedefinedtargetmodel.Manytoolsareavailablethatallowtransformationofalmostanydataformatanddatabase(e.g.CSV,Excel,relationaldatabases)toRDF.24

o Storeandpublish theRDFdata:ThegeneratedRDFdata is typicallystored inanRDFdatabase(triplestore)where itcanbeaccessedviaawebserverorqueriedatanSPARQLendpoint;thedata is also often published as a so called “RDF dump” (a RDF dataset made available fordownload).

o LinktootherRDFdataontheWeb:AccordingtotheLinkedDataprinciplespublishersshouldlinktootherdatasets to createanenrichedwebof LinkedData. Therefore relevant linking targetsneedtobeidentifiedwhichcanaddvalue(i.e.whererelationshipsexistbetweendata)andarewellmaintained. Publishersmay be aware of such datasets in their domain or search existingregistries(e.g.DataHub)toidentifyrelevantdatasets.Ifthereisarelevantdataset,thepublishermustdecidewhichproperties fromestablisheddomainor general LinkedDatavocabularies touseforthelinking.

o Describe,registerandpromotethedataset:ThepublisherofasetofLinkedDatashoulddescribethe dataset with metadata (including provenance, licensing, technical and other descriptiveinformation)whichcanbeattachedtothedataset.Itisalsogoodpracticetoregisterthedatasetin a domain data catalogue and general registries such as the DataHub. Furthermore thepublisher should announce the dataset via relevant mailing lists, newsletters etc. and inviteotherstoconsiderlinkingtothedataset.

There aremany introductory and advanced level guides available that describe how to generate,publish,linkanduseLinkedData:AsintroductorylevelguidesBauer&Kaltenböck(2012),Hyland&Villazón-Terrazas (2011)andW3C (2014)canbesuggested.Advanced“cookbooks”are theEUCLID

23OpenRefine,http://openrefine.org24W3Cwiki:ConvertertoRDF,http://www.w3.org/wiki/ConverterToRdf



curriculum25,Heath&Bizer(2011),Morganetal.(2014);NgongaNgomoetal.(2014),vanHooland&Verborgh(2014)andWoodetal.(2014).

Concerninguseful tools suchasRDF converters, LinkedDataeditors,RDFdatabases, etc. theW3Cwikiprovidesanextensive tooldirectory26.Someprojectsdescribeselectedtools they recommendfordifferent tasksof theLinkedData lifecycle, forexample, theprojects LATC (various tools)27andLOD2(mainlytoolsoftheprojectpartners)28.ButadoptersoftheLinkedDataapproachshouldseekadditionalexpertadviceonwhichtoolsareprovenandeffectivefortheirdataandcertaintasks.

3.6 Briefsummaryandrecommendations

Briefsummary

The term LinkedData refers to principles, standards and tools for the generation, publication andand linkingof structureddatabasedon theW3CResourceDescription Framework (RDF) familyofspecifications.

The basic concept of Linked Data has been defined by Tim Berners-Lee in an article published in2006.Thisconcepthelpedtore-orientateandchanneltheinitialgrandvisionoftheSemanticWebintoaproductivenewavenue.PreviouslytheresearchanddevelopmentcommunitypresentedtheSemanticWeb vision as a complex stack of standards and technologies. This stack seemedalways“under construction” and together with the difficult to comprehend Semantic Web terminologycreatedtheimpressionofanacademicactivitywithlittlerealworldimpact.

In 2010 Berners-Lee’s request for LinkedOpen Data aligned the Linked Data with the Open Datamovement.SincethenthequestforLinkedOpenData (LOD)hasbecomeparticularlystrong inthegovernmental/publicsectoraswellasinitiativesforculturalandscientificLOD.

TheLinkedDataprinciples includethatadatapublishershouldmakethedataresourcesaccessibleontheWebviaHTTPURIs(UniformResourceIdentifiers),whichuniquelyidentifytheresources,anduseRDFtospecifypropertiesofresourcesandofrelationsbetweenresources.InordertobeLinkedDataproper,thepublishersshouldalsolinktoURI-identifiedresourcesofotherproviders,henceaddtothe“webofdata”andenableuserstodiscoverrelatedinformation.AndtobeLinkedOpenDatathepublishermustprovidethedataunderanopenlicense(e.g.CreativeCommonsAttribution[CC-BY]orreleaseitintothePublicDomain).

The LinkedData approach allows opening up “data silos” to theWeb, interlink otherwise isolateddataresources,andenablere-useoftheinteroperabledataforvariouspurposes.Thelandscapeofarchaeological data is highly fragmented. Therefore Linked Data are seen as a way to interlinkdispersedandheterogeneousarchaeologicaldataand,basedon the interlinking,enablediscovery,accesstoandre-useofthedata.

Building semantic e-infrastructure and services for a specific domain such as archaeology requirescooperation between domain data producers/curators, aggregators and service providers.Cooperation is necessary not only for sharing datasets through a domain portal (i.e. theARIADNEdataportal),butalsotousecommonoralignedvocabularies(e.g.ontologies,thesauri)fordescribingthedatasothatitbecomesinteroperable.

25EUCLID-EducationalCurriculumfortheUsageofLinkedData,http://euclid-project.eu26W3Cwiki:Tools,http://www.w3.org/2001/sw/wiki/Tools27LATC-LODAroundTheClock(EU,FP7-ICT,9/2010-8/2012),http://latc-project.eu28LOD2-CreatingKnowledgeoutofInterlinkedData(EU,FP7-ICT,9/2010-8/2014),http://lod2.eu



In addition to the basic Linked Data principles there are also specific recommendations forvocabularies. Particularly important is re-using or extending wherever possible establishedvocabulariesbeforecreatinganewone.The rationale for re-use is thatdifferent resourceson theweb of Linked Data which are described with the same or mapped vocabulary terms becomeinterlinked. This makes it easier for applications to identify, process and integrate Linked Data.Moreover,re-useandextensionofexistingvocabulariescanlowervocabularydevelopmentcosts.

ItisalsorecommendedtoprovidemetadataforLinkedDataofdatasetsaswellasvocabularies.TheVocabularyofInterlinkedDatasets(VoiD)isoftenbeingusedforprovidingsuchmetadata.It isalsogoodpractice to register sets of LinkedData in a domain data catalogue and/or general registriessuchas theDataHub.Furthermore thepublisher shouldannounce thedatasetvia relevantmailinglists,newslettersetc.andinviteotherstoconsiderlinkingtothedataset.

Linked Data should not be published “just in case”. Rather publishers should consider the re-usepotential and intended or possible users of their data. As Linked Data consumers they need toaddress the question of which data of others they could link to. These questionsmake clear theimportance of joint initiatives for providing and interlinking datasets of certain domains such asarchaeology.

Recommendations

o UsetheLinkedDataapproachtogeneratesemanticallyenhancedandlinkedarchaeologicaldataresources.

o Participate in joint initiatives for providing and interlinking archaeological datasets as LinkedOpenData.

o Choose datasets which allow generating value if made openly available as Linked Data andconnectedwithotherdata,includinglinkingofthedatasetsbyothers.

o Re-useexistingLinkedDatavocabularieswhereverpossibleinordertoenableinteroperability.

o Describe the Linked Data with metadata, including provenance, licensing, technical and otherdescriptiveinformation.

o Registerthedatasetinadomaindatacatalogueand/orgeneralregistriessuchastheDataHub.Alsoannouncethedatasetviarelevantmailinglists,newslettersetc.andinviteotherstoconsiderlinkingtothedataset.



4 TheLinkedOpenDataCloud

ThischapterdescribeswhathasbeentermedtheLODCloudandisgenerallyillustratedwiththeLODCloud diagram of interlinked datasets. Some available figures for the state of the LOD Cloud arepresented and also some issues highlighted. Furthermore an overview of cultural heritage LODpresentontheLODClouddiagramandotherknownculturalheritageLOD,includingarchaeologicalLOD,isbeinggiven.

4.1 LODCloudfigures

The LinkedOpenData (LOD)Cloud is formedbydatasets that areopenly availableon theWeb inLinkedDataformatsandcontainlinkspointingatothersuchdatasets.ThelatestLODCloudfiguresandvisualizationhavebeenpublishedonlineinAugust2014(Schmachtenbergetal.2014a[statisticsonline],2014b[paper]).TheyarebasedoninformationcollectedthroughacrawloftheLinkedDataweb inApril2014.Thecrawl found1014datasetsofwhich569 (56%) linked toat leastoneotherdataset; the 569datasetswere connectedby in total 2909 link-sets. The remainingdatasetswereonlytargetsofRDFlinks,andthereforeattheperipheryofthe“cloud”,ortheywereisolated.Ofthe569coreLODClouddatasets374wereregisteredintheDataHub.29Thelatestcomparablefigurestothe ones reported by Schmachtenberg et al. (2014a/b) are based on the DataHub metadata ofdatasetsfromSeptember2011(Jentzschetal.2011)30.

BelowwesummarizesomeresultsofSchmachtenbergetal. (2014aand2014b,ofwhichthe lattercomparesthefiguresof2011and2014)whichgiveanimpressionoftheadoptionoftheLinkedDataprinciples:

o Increase indatasets:Therehasbeenasubstantial increase in identifieddatasets:2011:294LDdatasetsregisteredintheDataHub;2014:1014datasetsidentifiedthroughacrawlofthewebofLinkedData.With530datasetsthelargestgroupin2014wasthenewlyintroducedcategoryofsocial web/networking. These datasets describe people profiles and social relations amongstpeople.Among the established categories three showed a large growth in numberof dataset,Government (2011: 49; 2014: 183), Life Sciences (2011: 41; 2014: 83) and User-generatedcontent(2011:20;2014:48).

o Linkingofdatasets:445(43.89%)ofthe1014datasetsdidnotsetanyout-gowingRDFlinks,176(17.36%) linked to one other dataset, 106 (10.45%) to two datasets, 127 (12.52%) to 3-5datasets,81(7.99%)to6-10datasets,and79(7.79%)eventomorethan10datasets.

o AlesscentralizedLODCloud: In2014theweboflinkeddataappearedtobelesscentralized.In2011thecross-domainLinkedDataresourceDBpedia.orgclearlyoccupiedthecentreoftheLODCloud. In2014alsoGeoNameswasusedwidelyand therewere somecategory-specific linkinghubs(e.g.data.gov.uk inthecategoryGoverment).Most interconnectedwereresourcesofthecategoryPublications(e.g.RKBExplorerdatasets)andofthecategoryLifeSciences(e.g.Bio2RDFdatasets).

o Use of vocabularies: The 2014 survey discovered in total 649 vocabularies. 271 vocabularies(41.76%) were “non-proprietary”, defined as used by at least two datasets. Among these

29DataHub(OpenKnowledgeFoundation),http://datahub.io30StateoftheLODCloud,19/09/2011,http://lod-cloud.net/state/



vocabularies,RDFandRDFSaside,themostusedwereFOAF31(701datasetsusedit)andDublinCore32 (568 datasets used it). A special analysis showed that among the 378 “proprietary”vocabularies (defined as used by only one dataset) only 19.25% were fully and 8% partiallydereferencable; 72.75% had term URIs which were not dereferencable at all. One or moreproprietaryvocabularieswereusedby241datasets(23.17%ofthetotal).

o MetadataforsetsofLinkedData:For35.77%ofallsetsofLinkedDatain2014machine-readableprovenance and other metadata were provided (most often in Dublin Core, DCTerms orMetaVocab), about the same percentage than in 2011 (36.63%). Only about 8% providedmachine-readablelicensinginformation,mostlydc:license/dc:rightsandcc:license.HencelackofmetadataforsetsofLinkedDataremainsanissue.

4.2 (Mis-)readingtheLODdiagram

Intheyears2007-2011adiagramoftheLODCloudhasbeenproducedbasedondatasetsregisteredin the DataHub. The latest version of the diagram has been published in August 201433 and inadditiontotheDataHubinformationusestheresultsofacrawloftheLinkedDataWebinApril2014(Schmachtenberg et al. 2014a/b, as summarized above). The LOD Cloud diagram has grownenormously,toolargetopresentithere.

ThecriteriaforincludingadatasetintheLODClouddiagramare34:

o Theremustberesolvablehttp://(orhttps://)URIs.

o Theymustresolve,withorwithoutcontentnegotiation,toRDFdatainoneofthepopularRDFformats(RDFa,RDF/XML,Turtle,N-Triples).

o Thedatasetmustcontainatleast1000triples.

o Thedatasetmust be connected via RDF links to at least one other dataset in the diagram, byusingURIsfromthatdatasetorviceversa;atleast50linksarerequired.

o Access of the entire dataset must be possible via RDF crawling, an RDF dump or a SPARQLendpoint.

TheLODClouddiagramsthatsince2007havebeenproducedbasedonthesecriteriashowedsomelinkinghubs,butin2014therestillweremanyratherisolateddatasets(e.g.linkedtoonlyoneotherLinked Data resource). Yet the LOD Cloud diagrams have often been misleadingly referenced aspresentingacompact“webofdata”or“ahugeweb-scaleRDFgraph”(cf. thecritiquebyHogan&Gutierrez2014).Alsotheresearcherswhopublishedthe latest figuresontheLODCloudstate:“BysettingRDFlinks,dataprovidersconnecttheirdatasetsintoasingleglobaldatagraphwhichcanbenavigated by applications and enables the discovery of additional data by following RDF links”(Schmachtenbergetal.2014a).

Whatmustbeadded is that the“singleglobaldatagraph” ispatchy (asdescribedabove)andthatrelevant applications for end-users are hardly available. There are Linked Data browsers35 which,

31FOAF-Friend-of-a-Friend(definestermsfordescribingpersons,theiractivitiesandtheirrelationstoother

peopleandobject),http://xmlns.com/foaf/spec/32DublinCoreMetadataInitiative(DCMI)MetadataTerms,http://dublincore.org/documents/dcmi-terms/33TheLinkingOpenDataclouddiagram2014,byM.Schmachtenberg,C.Bizer,A.JentzschandR.Cyganiak,

availableat:http://lod-cloud.net34cf.TheLinkingOpenDataclouddiagram,http://lod-cloud.net



however, seem not to be in wider use, arguably because of a lack of interlinked data that arerelevantforusercommunities.Researchorienteddevelopershavecreatedsearchenginesbasedoncrawled and semantic Web Data (e.g. Sindice [service ended in 2014], Swoogle, Watson). Theseenginesareoflittleusefornon-experts.TheyserveasresearchtooltobetterunderstandtheLinkedData landscape. Research based on crawled Web data has become a specialty and is conductedaroundresourcessuchastheCommonCrawl36.The LOD Cloud is not a single entity but represents datasets of different providers that aremadeavailable in different ways (e.g. LD server, SPARQL endpoint, RDF dump) and often with lowreliability. For example, Buil-Aranda et al. (2013) found that of 427 public SPARQL endpointsregistered in the DataHub the providers of only one-third gave descriptive metadata. Half of theendpoints were off-line and only one third was available more than 99% of the time during amonitoringof27months;thesupportofSPARQLfeaturesandperformanceforgenericquerieswasvaried.PublicSPARQLendpointscouldformadistributed infrastructureforfederatedqueries37ofrelevantdataofdifferentsources(Rakhmawatietal.2013).Therebyviewsacrossthedifferentdatasetscouldbeprovided,allowingresearcherstoexplorethedata.ButthisdependsonreliablemaintenanceofthedatasetsandSPARQLendpointsbytheserviceproviders. Insteadofqueryingthe“singleglobalgraph” or just a number of LD datasets, the typical approach is to pull the data into one datarepository and run queries over this database. This approach is impractical for any but a smallnumber of datasets (or datasets of a small size), especially if only some interlinking between thedatasetsisofinterest.For intelligent searching, question answering and reasoning over Linked Data much more isnecessarythanprovidingSPARLendpointsorpullinganumberofdatasetsintoonegraphdatabase.One approach is “reason-able views” of LinkedDatawhich has been developed by researchers ofOntotextanddemonstratedwiththeFactForgeservice38(Kiryakovetal.2009;Damova2010;Simov& Kiryakov 2015). A reason-able view is constructed by assembling different datasets andvocabularies intoacompoundsetofLinkedData,producemappingsbetween instancedataof thedatasets, and create a single ontology for querying the compound dataset using SPARQL. Theontologyiscreatedbasedonmappingsbetweenthevocabulariesand/oranupper-levelontology,inthecaseofFactForge:PROTON39.Damova&Dannells(2011)illustratetheapproachwitha“museumreason-ableview”includingmappingsbetweenCIDOCCRMandPROTON,CIDOCCRMandSwedishOpenCulturalHeritage(K-samsök)40,andinformationoftheGothenburgCityMuseumtransformedtoRDF.AlsoexistingmappingsofDBPediaandGeoNamestoPROTONwereincluded.Areason-ableviewprovidesacontrolledenvironmentofintegrateddatasetstoexploitexistingandnewlycreatedsetsofLinkedData,reducedevelopmentcostsandrisksofunreliabledatasets.ThereisnocentralmanagementofLODCloud,theassumed“hugeweb-scaleRDFgraph”,but(some)areas forwhichacommunityofdevelopersproducesand interlinksrelevantresourcesandcreatesapplicationsforthepurposesoftheintendedend-users.InsuchcasesnetworkeffectsinthewebofLinkedDataarebeingachieved.Sucheffectsdonotresultautomaticallyfrommerelyputtingmore

35LODBrowserSwitch(offersasetofbrowsers),http://browse.semanticweb.org36CommonCrawl,http://commoncrawl.org37W3C(2013)Recommendation:SPARQL1.1FederatedQuery,21March2013,

http://www.w3.org/TR/sparql11-federated-query/38Ontotext:FactForge,http://ontotext.com/factforge-links/39Ontotext:PROTON,http://ontotext.com/products/proton/40SwedishOpenCulturalHeritage(K-samsök):http://www.ksamsok.se/in-english/



datasets into the LODcloud, actual interlinking is required togenerateawebof LinkedData.OneexampleofeffectivelinkingistheLinkedDatacommunityofthebio-medicalandlifesciences.Inthisarea the Bio2RDF41 project has created 35 Linked Data sets of existing databases and interlinkedsomeofthem.Anotherwell-curatedareaisLinkedDataofthelibrarycommunity.Culturalheritageor archaeology is not yet an area of densly interlinked information. So far a community ofcooperatingLODproducers,curatorsandintegratorshasnotemerged.

4.3 CulturalheritageintheLODCloud

ThelatestLODClouddiagram(August2014)providesan indicatorforthestateofculturalheritageLinkedData.SofaronlyfewculturalheritageLDdatasetsshowuponthediagram,andtheydonotformacloselylinkedwebofLD.Noneofthedatasetsconcernsarchaeologyspecifically.SomemoreculturalheritageLDsetsexist,alsoa fewarchaeologicaldatasets.But theydidnotconformto thecriteria for being included in the LODClouddiagram, e.g. the requirementof being connected viaRDFlinkswithatleastoneothercompliantdataset(seesectionabove).

Belowwefirstlisttheculturalheritagedatasetswhichconformtothecriteria,notincludingdatsetsof the library sector (e.g. Bibliothèque nationale de France [data.bnf.fr] or DeutscheNationalbibliothek[DNB]):

o EuropeanaLOD:mentionedinthefirstplacebecauseitisthelargestculturalheritageLDdataset(20millionrecords)andcomprisesofrecordsofmuseums,archivesandlibrariesacrossEurope42.

o Swedish Open Cultural Heritage (K-samsök): a web service that harvests metadata from thedatabasesofculturalheritageorganisationsinSwedenandallowscreatingLDbasedinformationservices43.

o Archives Hub Linked Data: the Archives Hub44 aggregates and allows searching acrossdescriptionsofarchivalcollectionsheldatover250institutionsintheUK(asearchoftheportalfor “archaeology” produces over 1000 hits). Linked Data of a sub-set of the aggregateddescriptionshasbeenproducedbytheLOCAHproject(2010-2011)45.

o British Museum - Semantic Web Collection Online: provides Linked Data access to the samecollection records as theMuseum’s web presented Collection Online; the data has also beenorganisedusingtheCIDOCCRM46.

o AmsterdamMuseum: has been the first museum in the Netherlands to convert its completemuseumcollectiondatabase(over70,000records)toRDF;thedata includes linkstotwoGetty

41Bio2RDF:LinkedDatafortheLifeSciences,http://bio2rdf.org42EuropeanaLinkedData,http://labs.europeana.eu/api/linked-open-data/introduction/;asearchonthe

Europeanawebsitefor“archaeology”showsthattheprovidersofmostrelatedcontentaretheSwedishNationalHeritageBoard(812,971items)andtheUKPortableAntiquitiesScheme(236,627).ARIADNEpartnersarealsopresent:GermanArchaeologicalInstitute/ARACHNE(183,683items),ArchaeologyDataService,UK(34,197)andDataArchivingandNetworkedServices,Netherlands(6456).

43SwedishOpenCulturalHeritage(K-samsök):http://www.ksamsok.se/in-english/;seealso:DataHub,http://datahub.io/dataset/swedish-open-cultural-heritage

44ArchivesHub,http://archiveshub.ac.uk45ArchivesHub–LOCAH,http://data.archiveshub.ac.uk46BritishMuseum-SemanticWebCollectionOnline,http://collection.britishmuseum.org



thesauri (AATNed [Dutch version] and ULAN), GeoNames, and DBPedia pages (De Boer et al.2012and2013)47.

o Art& Architecture Thesaurus (AAT) of theGetty Research Institute: The only cultural heritageKOS on the 2014 LOD diagram; meanwhile two other Getty KOSs have become available:Thesaurus of Geographic Names (TGN) and Union List of Artist Names (ULAN); the CulturalObjectsNameAuthority(CONA)wasexpectedtofollowinFall2015butseemstorequiremoreeffortthanexpected48.

Thesecond listbelowpresents furtherculturalheritageandarchaeologicaldatasets inLinkedDataformats that are registered in the DataHub or of which we know from searching various othersources. The list is certainly not comprehensive, because there have been quite some culturalheritageprojects that trialled the LinkedDataapproach,however thewhereaboutsof the createdLinkedDataareoftenunclear.TheLinkedDataresourceslistedbelowareroughlyorderedaccordingtotheirrelevanceinthecontextofourstudy:

o Archaeology Data Service (ADS): ADS Linked Open Data initially has been produced in theSTELLARprojectbyconvertingdatabasesandCSVfilestoRDF,usingtheCRM-EHontology;thisRDFdataisavailablefromaSPARQLendpoint49.Accordingtotheirannualreport2014/2015ADSnowalsohaveLODofdepositedprojectarchives, includingtheprojectsRomanAmphora50andColonisationofBritain(seeCripps2014forbackground);thenumberofLODtriplesin2015was2,531,302,upfrom680,500 inthepreviousreportingperiod(ADS2015:26).Notably,ADSalsoconsume LOD from external sources to populate own metadata (e.g. Ordnance Surveygeographicdata51).

o Data Archiving and Networked Services (DANS): DANSlabs has produced LOD of metadatarecordsofmore than25,000data sets stored in theDANS-EASYdigitalarchive,which includestheE-DepotforDutchArchaeology;thiswasdone2013inademonstrationproject,buttheLOD(withlittlecross-linking) isaccessibleviatheirSPARQLendpointunderanOpenDataCommonslicense52.

o CLAROS - TheWorld of Art on the SemanticWeb: the data of this international collaborationcomes frommajorClassicscollections, including fromARIADNEpartnerDAI; thedatahasbeenpreparedforasearchportalbasedonCIDOCCRMmodelling;thedataserviceismaintainedbytheUniversityofOxford’se-ResearchCentreandoffersaSPARQLendpoint53.

o Cultura Italia: providesmetadata of a number of Italian heritage institutions; offers a SPARQLendpointforthemetadata;alsothePICOthesaurusisavailablefordownload54.

47AmsterdamMuseuminEuropeanaDataModelRDF,http://semanticweb.cs.vu.nl/lod/am;seealso:DataHub,

http://datahub.io/dataset/amsterdam-museum-as-edm-lod48GettyVocabulariesLOD,http://vocab.getty.edu49ADSLinkedOpenData,http://data.archaeologydataservice.ac.uk;STELLARproject,

http://archaeologydataservice.ac.uk/research/stellar/50RomanAmphorae:adigitalresource(UniversityofSouthampton,2005;updated2014),

http://archaeologydataservice.ac.uk/archives/view/amphora_ahrb_2005/51OrdnanceSurvey(UK),http://data.ordnancesurvey.co.uk52DANSlabs:EASYMetadataasLinkedOpenDataDemo,http://dans-labs.github.io/easy-lod/53CLAROS:Data,http://data.clarosnet.org54CulturaItalia:Dati,http://dati.culturaitalia.it



o English Heritage Places: contains metadata for about 400,000 nationally important places asrecordedbyEnglishHeritage55;alsosevenEnglishHeritageandotherUKthesauriareregisteredintheDataHub,butforthosewerefertotheLDversionsproducedintheSENESCHALproject56.

o Pleiades: a gazetteer for ancient world studies operated by the Institute for the Study of theAncient World (USA)57; Pleiades URIs are used in the digital classics network Pelagios tointerconnect scholarly ancient world resources through the places they refer to; the Pelagiosprojectprovidesservicesandtoolstoallowscholarsannotate,aggregate,accessanddisplaytheplacereferences58.

o Nomisma:providesasLODanontologyfordescribingcoinsandseveralnumismaticsdatasetsoftheAmericanNumismaticSocietyandinstitutionsinEurope;aSPARQLendpointisavailable59.

o PortableAntiquitiesScheme:PASdataoffindsintheUKhasbeenlinkedtoLDresourcesoftheOrdnanceSurvey(nationalmappingservice),Pleiades(gazetteer),BritishMuseum,NomismaandDBpedia60(cf.Pett2014a/b).

o LinkedARC.net61: Frank Lynam (Trinity College Dublin), produced Linked Data of data ofexcavationsatPriniatikosPyrgos(Crete),modelledprimarilyusingCIDOCCRManditstypevalueslink to terms of the FISH Archaeological Objects Thesaurus, British Museum and Gettyvocabularies. The project is particularly interesting as it demonstrated the integration ofexcavationdataofAmericanandIrishgroupsofarchaeologists,applyingtheLocus-PailmethodofexcavationandMoLASsingle-contextmethodrespectively.

o MONDIS:adatasetaboutmonumentdamagesdevelopedintheCzechresearchprojectMONDIS;includestheirdiagnosticMonumentDamageOntology(Cacciotti&ValachJ.2015)62.

o MisMuseos.net:a“semanticcatalog”ofmuseumsinSpainandtheirinformationaboutartworksand artists63; the solution builds on the GNOSS social and semantic platform (Maturanaet al.2013).

o MuseiItaliani:alistofgeo-referencedmuseumsinItaly;thatformuseumcategoriesthedatasetlinkstoDBpediaandforplacestoGeoNames64.

o ReLoad-RepositoryforLinkedOpenArchivalData:aprojectoftheArchivioCentraledelloStato,Istituto per i Beni culturali dell’Emilia-Romagna and regesta.exe (2010-2013), the projectdevelopedontologies forarchivaldatasourcesandproducedaLODdatasetof severalarchivalinventories;ReLoadprovidesaSPARQLendpoint65.

55EnglishHeritagePlaces,DataHubinformation:http://datahub.io/dataset/englishheritage_places56HeritageData:Vocabularies,http://www.heritagedata.org/blog/vocabularies-provided/57Pleiades,http://pleiades.stoa.org58Pelagios,http://commons.pelagios.org59Nomisma,http://nomisma.org/datasets60PortableAntiquitiesScheme,http://finds.org.uk61Linkedarc.net,http://linkedarc.net;datasets,https://datahub.io/dataset/linkedarc62MONDISproject,http://www.mondis.cz;DataHubinformation:http://datahub.io/dataset?q=mondis63MisMuseos.net,DataHubinformation:http://datahub.io/dataset/mismuseos-gnoss64MuseiItaliani,http://www.linkedopendata.it/datasets/musei65ReLoad,http://labs.regesta.com/progettoReload/,seealsotheirprojectdescriptionfortheLODLAM2013

Summitchallenge,http://summit2013.lodlam.net/2012/12/01/challenge-entry-reload-repository-for-linked-open-archival-data/



SomeofthedatasetslistedabovemayshowuponthenextversionoftheLODClouddiagram,mostlikelythosewhicharemaintainedandemployedbyadedicatedgroupofdevelopersandusers liketheNomismaontologyanddatasetsandthePleiadesgazetteer,forinstance.

TheArt&ArchitectureThesaurus(AAT)asalinkinghub

Alreadyonthe2014LODClouddiagramwastheArt&ArchitectureThesaurus(AAT)whichtheGettyResearch Institute in February 2014 released as LOD. The multilingual AAT contains over 40,000concepts and over 350,000 terms for describing objects of visual art, architecture, othermaterialheritage, archaeology, conservation, archivalmaterials, etc. The AAT has the potential to becomeoneof the core linkinghubs for cultural heritage information in the LinkedOpenDataCloud. In asurveyonLinkedDataoftheAthenaPlusprojecthalfofthe24projectpartnerssaidthey intendtolink to theAAT and otherGetty thesauriwhen they are available as LOD (AthenaPlus 2013b: 10).When the AAT was released as LOD, among the initiatives that started using it was Europeana.Europeana partnerswho already use AAT termswere invited to re-submit theirmetadata so thattheiroldAAT term labels (providedasa simple text string)couldbeautomatically replacedby thenewAATURIs(Charles&Devarenne2014).ThisenableslinkingtoinformationofothersonthewebwhousetheseURIs.ThisisalsopossibleifdataprovidersmaptheirlocalvocabularytotheAAT.InARIADNEthedataprovidersmappedtermsofvocabularies(e.g.nationalthesauriorowntermlists)which theyuse for their datasetmetadata to appropriate termsof theAAT, using SKOSmappings(e.g.skos:exactMatch,skos:closeMatchandothers).


Briefsummary

TheLinkedOpenDataCloud is formedbydatasets thatareopenlyavailableon theWeb inLinkedDataformatsandcontainlinkspointingatothersuchdatasets.OnetaskoftheARIADNEprojectistopromote the emergence of a web of interlinked archaeological datasets which comply with theLinkedOpenData(LOD)principles.ItisanticipatedthatthiswebofarchaeologicalLODwillbecomepartofthewiderLODCloudandinterlinkedwithrelatedotherdataresources.

ThelatestLODClouddiagram(2014)includesonlyfewsetsofculturalheritageLODandtheydonotform a closely linkedweb of LinkedData. None of the datasets concerns archaeology specifically.Somemore sets of cultural heritage LinkedData sets exist, also a few archaeological, but in 2014they did not conform to the criteria for being included in the LOD Cloud diagram (e.g. therequirementofbeingconnectedviaRDFlinkswithatleastoneothercompliantdataset).

MaybethenextversionoftheLODClouddiagramwillcontainsomeoftheearlierandmorerecentsetsofarchaeologicalLinkedOpenData.Hopefullythiswillincludesomerelevantvocabularieswhichrecentlyhavebeen transformed to LinkedData in SKOS format. In 2014 theonly cultural heritagevocabularyonthediagramwastheArt&ArchitectureThesaurus(AAT),whichhasthepotential tobecomeoneofthecorelinkinghubsforculturalheritageinformationintheLODCloud.The LOD Cloud is not a single entity but represents datasets of different providers that aremadeavailableindifferentways(e.g.LDserver,SPARQLendpoint,RDFdump)andtheresourcesareoftenunreliable, e.g.many SPARQL endpoints are off-line. There is no centralmanagement and qualitycontrol of the LODCloud.Webs of reliable and richly interlinked datasets are only presentwherethere isacommunityofLinkedDataproducersandcurators(e.g. intheareasofbio-medical& lifesciencesorlibraries).



CulturalheritageorarchaeologyisnotyetanareaofdenslyinterlinkedandreliableLODresources;sofaracommunityofcooperatingLODproducersandcuratorshasnotemerged.Targetedactivitiestofosterandsupportfurtherpublicationand interlinkingofdatasetsarerequiredsothatawebofarchaeological,culturalheritageandotherrelevantdatawillemergewithintheoverallLinkedOpenDataCloud.

Recommendations

o Encouragearchaeological institutionsandrepositoriestopublishthemetadataoftheirdatasets(collections,databases)asLinkedOpenData;alsopromotepublicationofdomainandproprietaryvocabulariesofinstitutionsasLOD.

o Foster the formation of a community of archaeological LOD producers and curators whogenerate,publishandinterlinkLOD,includinglinking/mappingbetweenvocabularies.



5 AdoptionoftheLinkedDataapproachinarchaeology

Sinceabout10years theSemanticWeb/ LinkedData standards,methodsand toolshavebecomemorematureandapplicable.CulturalheritageinstitutionshavebeenamongtheleadingadoptersoftheLinkedDataapproach,mainlytobetterinterlinkdomainresourcesand,insomecases,toenrichtheironlineinformationwithinformationofpopularresourcessuchasDBpedia/Wikipediacontent.With regard toLinkedDataofarchaeologicalprojectarchivesanddatabases therehavebeenonlyfewprojects,witharguablylimitedrecognitionbythewiderarchaeologicalresearchcommunity.Atthe same time, there has been a boom in LinkedData projects in the AncientWorld and Classicsresearchcommunity.Thischapterdescribesandaimstoexplainthissituationingreaterdetail.

5.1 Adoptionbyculturalheritageinstitutions

Institutionsoftheculturalheritagesector,particularlylibrariesandmuseums,areamongtheleadingadoptersof the LinkedData approach. In an international survey for institutional implementersofLinkedDataservicesbyOCLCResearchin2015,seventy-oneinstitutionsfrom16countries(45%USA)reported in total 168 Linked Data projects (Smith-Yoshimura 2016). The survey had a focus onlibraries, but also some other organisations participated (e.g. American Numismatic Society, TheBritishMuseum,EuropeanaFoundation). Two-thirdsof theprojectswere completed (i.e. a serviceimplemented).

In the area of museums one pioneering project was Finnish Museums on the Semantic Web(Hyvönen et al. 2002)66, followed by many others, in recent years for example the AmsterdamMuseum (De Boer et al. 2012 and 2013)67, British Museum68, Peter the Great Museum ofAnthropology andEthnography in St. Petersburg (Ivanov2011), RussianMuseum in St. Petersburg(Mouromtsevetal.2015)andSmithsonianAmericanArtMuseum(Szekelyetal.2013).69

Archives appear to be less advanced in the application of LinkedData. Their initial steps focus onbringing legacy finding aids onlinewhile providing access to the archival records andmaterial stilloften requires much digitisation work. In recent years there has been some progress instandardisation thatwill help inmoving towards LinkedData. For example, efforts by the ExpertsGrouponArchivalDescription (EGAD, since2012) tomake theEncodedArchivalDescription (EAD,2002) standard more data-centric in EAD3 (2015) and better connect it with Encoded ArchivalContext–CorporateBodies,PersonsandFamilies (EAC-CPF,2010)andotherstandards70 (Gueguenetal.2013;Pittietal.2014).

Currently the archive community seeks to establish guidelines for structuring archival LinkedDataresourceswiththenewstandards,buildsupportforeditingandpublicationintoarchivaltools(e.g.easeaddingidentifiersofauthorities),andderivegoodpracticefromtheexperienceoffirstprojectsin the field (Gracy&Lambert2014;Gracy2015).ExamplesofpioneerprojectsareLOCAH-Linked

66TheSemanticComputingResearchGroup(SeCo)atAaltoUniversity(Finland),wholedtheproject,continues

tobealeaderinLinkedDataapplicationsforculturalheritageresources,http://seco.cs.aalto.fi67AmsterdamMuseumasLinkedOpenDataintheEuropeanaDataModelAmsterdamMuseum,

http://semanticweb.cs.vu.nl/lod/am68BritishMuseum-SemanticWebCollectionOnline,http://collection.britishmuseum.org69SomeotherexamplesarelistedontheMuseumsandtheMachine-processableWebwiki,http://museum-

api.pbworks.com/w/page/21933420/Museum%C2%A0APIs70EncodedArchivalDescription(officialsite),http://www.loc.gov/ead/



ArchivesandLinkingLives(2010-2012)71(Stevenson2012)andReLoad-RepositoryforLinkedOpenArchival Data (2010-2013)72 (Mazzini & Ricci 2011). The LiAM - Linked Archival Metadata project(2012-2013)73 provides a guidebook that helps applying Linked Data approaches to archivaldescription(Morganetal.2014).

While there exists no comprehensive overview of cultural heritage Linked Data projects, studieswhichdescribeseveralexamples(e.g.Edelsteinetal.2013a/b)typicallydonotincludearchaeologicalprojects. But there is a significant difference between cultural heritage institutions and researchorganisationsandprojects.Culturalheritageinstitutionssuchaslibraries,archivesandmuseumsaremotivated by a service ethos, the mission to make information about heritage readily available.Researchersareprimarilyinterestedtopublishresearchresults,whilestilllittleacademicrewardcanbegainedfromsharingthedataunderlyingtheresults.ThereforeLinkedDataoflegacydatasetsmaybe easier to promote than data of current research, where first the objective of “open data” ingeneralneedstobeaddressed(ARIADNE2015e:chapter4;Carver&Lang2013).

5.2 Lowuptakeforarchaeologicalresearchdata

In theculturalheritagesector therehavebeen initiativespromotingtheLinkedDataapproach, forexample, LOD-LAM, the International LOD in Libraries, Archives, and Museums Summit (since2011)74,ortheLinkedHeritageproject75whichdisseminatedguidanceforLinkedDatatomuseumsinEurope.76Inthefieldofarchaeologicalresearchtherewerenosuchinitiativesoronlyatsmallscale,forexample,sessionsatCAAconferencesornationalthematicworkshops.Butpromotionalactivities,particularly at the national level, are important to reach archaeological institutes and researchgroupsandmakethemawareoftheLinkedDataapproach.Forexample, inFrancetheConsortiumMASA77aimstoprovidearchaeologistswithvocabulariesandtoolstoimprovetheinteroperabilityoftheirdataviaLinkedDatastandards.MASAisoneofthetenconsortiumoftheHUMA-NUMresearchinfrastructurewhichfocusonparticularresourcesandfieldsof(digital)humanitiesresearch78.

InARIADNEaLinkedDataSpecial InterestGroup (SIG)79hasbeen formed thatactsasan interfacewiththewiderLinkedDatacommunity,communicatingdevelopmentsbetweenthecommunityandARIADNE(andviceversa), looking forsynergy,andrelevantcommonusecases.Participantsof thefirstmeetingoftheARIADNELinkedDataSIG(2013)notedastill lowuptakeorevenawarenessof71LOCAH-LinkedArchivesandLinkingLives(UK,2010-2012,ArchivesHub),http://locah.archiveshub.ac.uk72ReLoad-RepositoryforLinkedOpenArchivalData(Italy,2010-2013,ArchivioCentraledelloStato,Istituto

periBeniculturalidell’Emilia-Romagnaandregesta.exe),http://labs.regesta.com/progettoReload/;seealsotheirprojectdescriptionfortheLODLAM2013summit(ReLoad2013).

73LiAM-LinkedArchivalMetadataproject(USA,2012-2013,ledbyTuftsUniversity,DigitalCollectionsandArchives),http://sites.tufts.edu/liam/

74LOD-LAM,http://lodlam.net75LinkedHeritage(EU,ICT-PSP,2011-2013),http://www.linkedheritage.eu76AstrongimpacthavealsohadtheculturalheritageaggregationprojectssuchasCulturaItalia

(http://dati.culturaitalia.it);SwedishOpenCulturalHeritage(K-samsök,http://www.ksamsok.se/in-english/),andofcourseEuropeana,whichhaspublishedoneofthelargestLinkedDatasetscomprisingrecordsofmuseums,archivesandlibrariesacrossEurope(http://labs.europeana.eu/api/linked-open-data/introduction/).

77MASA-MémoiredesArchéologuesetdesSitesArchéologiques,http://masa.hypotheses.org78HUMA-NUM:Consortiums,http://www.huma-num.fr/consortiums79ARIADNELinkedDataSIG,http://www.ariadne-infrastructure.eu/Community/Special-Interest-

Groups/Linked-Data



theLinkedDataapproachbyarchaeologicalresearchandotherorganisations.Theparticipantssawaclear need of raising awareness of advantages offered by Linked Data and promoting furtheradoption in the sector. Furthermore, to leverage the creation and interlinking of Linked Dataresources,practicalguidanceandeasytousetoolsarenecessary.

In the secondmeeting of the ARIADNE Linked Data SIG (2014), Leif Isaksen, the chair of the CAASemantic SIG80, characterized the current phase of archaeological Linked Data as “a period ofexperimentation”.GroupmembersexpectedthatfromthisexperimentationsomeprojectswillpavethewaytoabroaderadoptionandincreasingutilityofLinkedDatainarchaeology.

The requirements for a wider uptake recognised by the ARIADNE Linked Data SIG are alsoemphasisedby thecommunity thataims to interlink informationabout theancientworld. In2012the3-dayLinkedAncientWorldDataInstitutemeeting(LAWDI2012)broughttogetherprojectsandinterestednewusersinthisfield.Themeetingreportnotes:“EssentiallyallLAWDIparticipantswereeagertoshowresourcesthatprovidestableURIsortoaskforadviceonwhatiscurrentlyavailable.ButboththeparticipantsinandorganizersofLAWDIrecognizetheneedtotakeactivestepstogrowthe number of high-quality digital resources. That will require ongoing outreach as well as clearexamplesofhowLinkedOpenDatabenefitsbothcreatorsandusers” (Elliott,Heath&Muccigrosso2012:45).

FromtheLinkedAncientWorldDataInstitute(LAWDI)meetingsin2012and2013acollectionof30articlesoriginatedwhich illustratestheadoptionoftheLinkedDataapproach intheAncientWorldresearch community and what it takes to move from concept to actual implementation andoperation (Elliott, Heath&Muccigrosso 2014). The papers cover awide range of cultural objects,topics and information resources including, among others, cuneiform tablets, epigraphy,numismatics,prosopography(informationaboutpeople),ancientandclassicalliterature,publicationof bibliographies and reviews, location/mapping services, historical periodization, integration ofhistorical-geographicinformation,andmore.

5.3 TheAncientWorldresearchcommunityasafront-runner

At the “Linked Pasts” colloquium, which was organised by the Pelagios project at King’s CollegeLondon (20-21 July 2015), one topicwas the importance to demonstrate benefits of using LinkedOpenData.LODdevelopers inresearchfieldsofancienthistoryandclassicswererecognisedbeingclosertothisgoalthanearlyadopters inarchaeology.AssummarizedinanarticleontheARIADNEwebsite:“OfmostinteresttoARIADNEwerethereasonsClassicshasbeenmoresuccessfulthanothercultural heritage domains (i.e. archaeology generally) at successfully implementing LOD. This wasstated as primarily down to a lack of resources, heterogeneity of data, and (therefore) difficultydemonstratingclearbenefits”(ARIADNE2015d).WhenweaskwhysomefieldsofAncientWorldandClassicsresearcharemoreadvancedthanArchaeologywithregardtoLinkedData,theheterogeneityofdatainarchaeologicalprojectarchivesanddatabasesindeedisamajorfactor.

Advantageofspecialties

Whilearchaeologistsunearthanddocumentalargevarietyofbuiltstructures,culturalartefactsandbiologicalremains,relatedAncientWorldandClassicsresearchspecialtiestypicallyfocusononetypeofartefactssuchasinscriptions(epigraphy),coins(numismatics),ceramics,andothers.Consequentlyin these (smaller) research communities it is easier to establish and promote the use of common

80CAASemanticSIG,https://groups.google.com/forum/#!forum/caa-semantic-sig



descriptionstandards.Thesestandardsareappliedtodatabasesofartefactcollections,whichhaveoften been created (at least in part) from finds of archaeological excavations. The differencegenerallyisthatinarchaeologythebasicunitofresearchandanalysisisthearchaeologicalsite,whileresearchinspecialitiesofAncientWorldandClassicsbuildsoncollectionsor, inthecaseoftexts,acorpus.

OneleadingexampleamongthespecialtiesistheinternationalNomisma81collaboration(since2010)thatdevelopsdescriptionstandardsforcoins(e.g.theNomismaOntologywhichprovidesstableURIsfor numismatic concepts and entities), produces LinkedData sets ofmajor collections, and sharesthem under open licenses. One reference implementation is Online Coins of the Roman Empire(OCRE)82oftheAmericanNumismaticSociety(Gruberetal.2013;Meadows&Gruber2014).

TheontologyandLinkedOpenDatamethodologiesestablishedbyNomismaareemployedbyseveralother numismatics resources, for example, Antike Fundmünzen Europa83, a web-based coinsdatabasedevelopedby theRomano-GermanicCommissionof theGermanArchaeological Institute(Tolle&Wigg-Wolf2016).TheCommissionalsocoordinatestheEuropeanCoinFindNetwork-ECFNandseveraljointmeetingsofECFNandNomismahavebeenorganised84.

ConcerningpotterydatasetstheKerameikos85 initiativefollows lessons learned inthedevelopmentofNomismaandaimstodevelopa thesaurus thatdefinesdomainconceptswithURIsandRDF forrepresentingandsharingpotterydataacrossdisparatesystems.The initiativehasbeen introducedwithapaperattheCAA2014conferenceinParisthatdemonstratesthepotential(Gruber&Smith2015),followedbyaroundtableonLODappliedtopotterydatabasesattheCAA2015conferenceinSiena (Gruber et al. 2015). Initially Kerameikos focuses on concepts within Greek black- and red-figurepottery,tobeextendedtootherfieldsofpotterystudies.SeealsothecasestudypresentedbyThiery(2014)onaLODapproachtosimianware,linkingpotters,potsandplaces.

Anotherbroadfieldofresearchisinscriptions(epigraphy),wheretheEuropeanaNetworkofAncientGreek and Latin Epigraphy (EAGLE)86 project has achieved a substantial advance (Casarosa et al.2014;Liuzzo2014and2016).ThisincludesaconceptualandametadatamodelbasedonCIDOCCRMandTEI/EpiDoc,respectively(EAGLE2015),andasetofvocabulariesforclassicalepigraphyinSKOSformat87.

Coins, pottery and inscriptions are but three examples chosen because they concern materialartefactsfamiliartoarchaeologists.OtherexamplesofLODorientedinitiativesconcernthedomainofancientandclassical texts. Forexample, theStandards forNetworkingAncientProsopographies(SNAP)88 project defines annotation conventions and builds a single virtual authority list forreferencingancientpeople,broughttogetherfromdifferentauthoritativelistsofpersonsandnames.

81Nomisma,http://nomisma.org82OnlineCoinsoftheRomanEmpire(OCRE),http://numismatics.org/ocre/83AntikeFundmünzeninEuropa(AFE),http://afe.fundmuenzen.eu84EuropeanCoinFindNetwork(ECFN),http://www.ecfn.fundmuenzen.eu85Kerameikos,http://kerameikos.org86EuropeanaNetworkofAncientGreekandLatinEpigraphy-EAGLE(EU,ICT-PSP,4/2013-3/2016),

http://www.eagle-network.eu87EAGLEvocabularies(Material,Typeofinscription,Executiontechnique,Objecttype,Decoration,Dating

criteria,Stateofpreservation),http://www.eagle-network.eu/resources/vocabularies/88StandardsforNetworkingAncientProsopographies–SNAP(UKAHRCfundedproject,2014-2015),

http://snapdrgn.net



AfocusoncommondescriptionstandardsforcertaintypesofAncientWorldartefactsandtextsdoesof course not mean ignoring their relations with other subject areas and common issues. As the“Linked Ancient World Data: Relating the Past” panel at the Digital Humanities 2016 conferenceexplains, theseprojects“arealso concernedwith issues far beyond their primary subject area: theinteroperabilityofbibliographicalreferences,citationsofancientsources,encodingofdateandtime,events and actors, material objects and their curatorial history all contribute to the study andunderstandingoftheancientworld(andmutatismutandisofanyother).Allalsorecognisethatthereis no firm demarcation between the cultures of the Mediterranean in the classical period, norbetween the worlds and cultures bordering them in time and space” (Linked AncientWorld Data2016).

Important to note is that all LinkedData effortsmentioned are about artefacts and texts,while alargesegmentofarchaeologicalresearchconcernsbiologicalremainsofhumans,animalsandplants.However, biological vocabularies are not developed by archaeologists, but by taxonomists (withregard tospeciesnames)89,Biodiversity InformationStandards (TWDG)90,whodevelopLifeScienceIdentifiers (LSID) and vocabularies for biodiversity information, and expert groups that producerelevantbiologicalontologieswhicharesharedviatheBioPortal91.Whileauthoritativespeciesnamesare widely used by archaeobotanists and zooarchaeologists, other standards such as biologicalontologiesseemtobeemployedseldom. Indeed,wefoundonlyexamplewheresuchanontology,theUberAnatomyOntology (UBERON)92hasbeenused ina zooarchaeological LinkedDataproject(Kansaetal.2014;Whitcher-Kansa2015).

Pelagiosasacommonplatform

The strongest impression of the Ancient World research community being a front-runner inhumanities LOD comes from Pelagios93, which since 2011 supports connecting various scholarlyresources through the places and other geographic entities they refer to. Pelagios is a looseconfederationofmanyorganisationsandprojects thathaveagreed touse for such references theOpenAnnotation94RDFvocabularyandURIsofgazetteersoftheancientworldgeography,inprimisPleiades95 but also others (e.g. iDAI.gazetteer96, Digital Atlas of the Roman Empire97, Vici.org98 andothers).Amongthecurrently21datasetcontributorsofPelagiosaretheARIADNEpartnersGermanArchaeological Institute(iDAI.objectsdatabasewith87,735referencesconcerning5363places)andFastiOnline(with686referencesconcerning256places)99.

Pelagiosaggregatestheannotations,whicharehostedbythedataproviders(oftenintheformofanRDF dump), andmakes them available through amap-based search interface and an API so that

89AmajorintegratorinthisfieldistheCatalogueofLife,http://www.catalogueoflife.org90TDWG-BiodiversityInformationStandards,http://www.tdwg.org91BioPortal(USNationalCenterforBiomedicalOntology),https://bioportal.bioontology.org92UBERON-UberAnatomyOntology,http://uberon.org93Pelagios,http://commons.pelagios.org94OpenAnnotationCollaboration,http://www.openannotation.org95Pleiades,http://pleiades.stoa.org96iDAI.gazetteer(GermanArchaeologicalInstitute),http://gazetteer.dainst.org97DigitalAtlasoftheRomanEmpire(DepartmentofArchaeologyandAncientHistory,LundUniversity,

Sweden),http://dare.ht.lu.se98Vici.org-ArchaeologicalAtlasofAntiquity(community-basedgazetteer),http://vici.org99Pelagios:Datasets,http://pelagios.org/peripleo/pages/datasets



developerscanbuildonthedata.TheannotationplatformRecogitoaids theprocessof identifyingplacesreferredtoinindividualdigitaltextsandmapsandlinkingthemtoagazetteer,supportedbyanautomatedsuggestionsystem(Simonetal.2015).CurrentlyindevelopmentisPeripleo,atooltoexplorethegrowingpoolofdataasawholeandtoprogressively filteranddrilldownto individualrecords(Simonetal.2016).

Isaksenetal.(2014)addressseveralfactorswhichdeterminedthesuccessofthePelagiosinitiative.AmongthemostimportantarguablyarethelightweightLinkedDataapproach,focusongeographicalreferences as the most common feature of the various data resources, quick demonstration ofbenefits from associating contributors’ data, and the sustained funding by the AndrewW.MellonFoundation(since2013,currentlybyagrantuntil2018100).Buttheyalsonote,“weareatthetipoftheicebergeveninthiscaseastheoverwhelmingmajorityofclassicistsandclassicalarchaeologistshaveneverheardofLinkedOpenData”(Isaksenetal.2014).

In summary,major factors that contribute to an advancedpositionof theAncientWorld researchcommunityintheapplicationoftheLinkedDataapproachare:a)therearegroupswhodevelopandpromotedescriptionstandards incertainspecialities,andb)there isacommonplatform(Pelagios)that allows linking of information based on a light-weight approach. Archaeological projects canbenefitfromthisdevelopment,forexample,usetheNomismadescriptionstandardsforcoinfinds.

100Initialfundingin2011-2012byJISC(UK)andgrantsforspecialprojectsin2014-2015byAHRC(UK)andOpen

KnowledgeFoundation.




Briefsummary

Intheareasaddressedbythisstudy,culturalheritageinstitutionsareamongtheleadingadoptersoftheLinkedDataapproach.TheAncientWorldandClassicsresearchcommunity isafront-runnerofuptake on the research side, while there have been only few projects for Linked Data ofarchaeologicalresearchdata.

Thissituationisduetoconsiderabledifferencesbetweenculturalheritageinstitutionsandresearchprojects, and between projects in different domains of research. For cultural heritage institutionssuch as a libraries, archives andmuseums adoption of LinkedData is in linewith theirmission tomake information about heritage readily available and relevant to different user groups, includingresearchers.AdoptionhasalsobeenpromotedbyinitiativessuchasLOD-LAM,theInternationalLODin Libraries, Archives, and Museums Summit (since 2011). In the field of archaeological researchtherewere no such initiatives or only at small scale, for example sessions at CAA conferences ornational thematic workshops. But promotional activities, particularly at the national level, areimportanttoreacharchaeologicalinstitutesandresearchgroupsandmakethemawareoftheLinkedDataapproach.

AdoptionintheAncientWorldandClassicsresearchcommunityisbeingdrivenbyspecialitiessuchasnumismaticsandepigraphy,wherethereareinitiativestoestablishcommondescriptionstandardsbasedonLinkedDataprinciples.Thegoalhereistoenableannotationandinterlinkingofinformationof special collections or corpora for research purposes. The focus on certain types of artefacts(inscriptions,coins,ceramicsandothers)provideclearadvantageswithregardtothepromotionofthe Linked Data approach within and among the relatively small research communities of thespecialities.

AgooddealoftherecognitionoftheAncientWorldandClassicsresearchcommunitybeingafront-runnerinLinkedDataalsostemsfromthePelagiosinitiative.Pelagiosprovidesacommonplatformand tools for annotating and connecting various scholarly resources based on place references.Pelagios clearly demonstrates benefits of contributing and associating data of the differentcontributorsbasedonalight-weightLinkedDataapproach.

Archaeologypresentsamoredifficultsituation, inthatthebasicunitofresearch is thesite,wherearchaeologists unearth and document a large variety of built structures, cultural artefacts andbiologicalmaterial. The heterogeneity of the archaeological data and the site as focus of analysispresentasituationwherethebenefitsofLinkedData,whichwouldrequiresemanticannotationofthevarietyofdifferentdatawithcommonvocabularies,arenotapparent.ThereforeadoptionoftheLinkedDataapproachcanbehardly foundatthe levelof individualarchaeologicalexcavationsandother fieldwork, but, in a few cases, community-level data repositories and databases of researchinstitutes.Repositoriesanddatabases,notindividualprojects,shouldalsoinnextyearsbetheprimetargetwhenpromotingtheLinkedDataapproach.

Allproponentsof theLinkedDataapproach, including theARIADNELinkedDataSIGaswellas thedirectorsofthePelagiosinitiative,agreethatmuchmoreneedstobedonetoraiseawarenessoftheapproach,promoteuptake,andprovidepracticalguidanceandeasytousetoolsforthegeneration,publicationandinterlinkingofLinkedData.



Recommendations

o MoreneedstobedonetoraiseawarenessandpromoteuptakeoftheLinkedDataapproachforarchaeological researchdata. Inaddition to sessionsat international conferences, promote theapproachtostakeholderssuchasarchaeologicalinstitutesatthenationallevel.

o Theprimetargetwhenpromotingtheapproachshouldbecommunity-leveldatarepositoriesanddatabasesofresearchinstitutes(notindividualprojects).

o To drive uptake provision of practical guidance and easy to use tools for the generation,publicationandinterlinkingofLinkedDataisnecessary.

o Promotetheuseofestablishedandemergingsemanticdescriptionandannotationstandardsforartefactssuchascoins,inscriptions,ceramicsandothers;forbiologicalremainsofplants,animalsand humans suggest using available relevant biological vocabularies (e.g. authoritative speciestaxons,lifescienceontologies,andothers).

o ContributetothePelagiosplatform(whereappropriate)oraimtoestablishsimilarhigh-visibilitydatalinkingprojectsforarchaeologicalresearchdata.



6 RequirementsforwideruptakeoftheLinkedDataapproach

Linked Open Data (LOD) allow for semantic interoperability of dispersed and heterogeneous dataresources.DespitethispotentialLODisnotproducedandappliedyetbymanyresearchinstitutionsandprojectsinthearchaeologicalsector.ThesectionsofthischapteraddressdifferentrequirementsandapproachesforfosteringawideruptakeoftheLinkedDataapproachforarchaeologicalresearchdata. The aim is to present the current state with regard to impediments, potential drivers andexemplaryprojects,andforeachareaofidentifiedrequirementsprovidepracticalrecommendationsforLinkedDatadevelopersandotherstakeholders.

6.1 RaiseawarenessofLinkedData

LinkedDataenableinteroperabilityofdispersedandheterogeneousinformationresources,allowingthe resources to become better discoverable, accessible and re-useable. In a fragmented datalandscapeaspresentinthesectorofarchaeologythisissubstantialvalueproposition.Indeed,inanARIADNEonlinesurveyontopoftheexpectationsofabout500researchers,researchdirectorsandother respondents fromadataportalwere cross-searchingofdataarchiveswith innovative,morepowerfulsearchmechanisms(ARIADNE2014a:114,about500respondents).

But such expectations are not necessarily associated with capabilities offered by Linked Data.Therefore the gap between advantages expected from advanced data services and “buy in” andsupportoftheresearchcommunityforLinkedDatamustbeclosedbytargetedactions.Thissectionaddresses the situation of a highly fragmented landscape of archaeological data, presents someavailable results on the awareness of Linked Data by cultural heritage organisations andarchaeologists,andsuggestswhomtoconsiderasprioritytargetgroupsforLinkedDatainitiatives.

6.1.1 Fragmentationofarchaeologicaldata

TheARIADNE“FirstReportonUsers’Needs”(ARIADNE2014a)identifiedmajorgeneralfactorsthatimpedetheuptakeoftheLinkedDataapproachinthedomainofarchaeologicalresearch.Theresultsof the literaturereview,pilot interviewsandonlinesurveymadeclearthatthearchaeologicaldatalandscape is characterizedbyhigh fragmentationdue to several factors.These factors include,butarenotlimitedto

- diverse organisational settings (research institutes, heritage management agencies, museumsandothers)inwhichdataarecollectedandmanaged,

- datamanagementpractices thatarepredominantly focusedon individualprojects, rather thananinstitutionalordomainorientedperspective(e.g.“projectarchives”,oneperexcavationsite,storedonafileservers,etc.),

- alowlevelofopensharingofresearchdata,duetolackofrecognitionandrewardsformakingthedataavailable,theadditionalworkeffortfordocumentingdatasetsforproperarchiving,andlackofcommunityarchivesinmanycountries.

Thesituationdoesnotpresentfavourableconditionsfortheintegrationandlinkingofarchaeologicaldatasetsthroughdatae-infrastructuressuchasARIADNE.ThereforeARIADNEencouragesinitiativesto establish state-of-the-art community-level data archives in countrieswhere they aremissing atpresent.Thissuggestionis in linewiththedevelopmentthatresearchfunders increasinglydemand



data management & access plans with the goal to make the generated research data openlyaccessiblethroughdigitalarchives(opendatamandates).

Research projects will have to think about data management from the start, including where todeposit theirdata, requiredmetadata,and licensingagreements.Alsosomescientific journalsnowrequire a data availability statement, i.e. that the data which underpins published research isavailableinanaccessiblearchive.HoweverwithregardtopromotingarchaeologicalLinkedDatatheprimaryfocusmustnotnecessarilybeindividualresearchers,researchgroupsandprojects.Becausedata produced by projects will increasingly be deposited in accessible data archives, according tosectorstandardswithregardtometadataandvocabularies.

6.1.2 CurrentawarenessofLinkedData

Resultsforculturalheritageorganisations

It is worthwhile having an indication of the current state of awareness and knowledge of LinkedOpen Data (LOD) at cultural heritage organisations, some of which may curate archaeologicalartefacts among other objects and content. The AthenaPlus project101 conducted a survey amongpartnersandotherorganisationsabouttheirawarenessofLODandexistinginitiatives,howtheygetinformation about LOD, and if they already use LOD (AthenaPlus 2013b). 28 questionnaires werereturned by respondents of organisations located in 16 EU countries. The respondents worked atmuseums, libraries, archives, data aggregators and other organisations, including ministries,governmentalagencies,universityresearchcentresandITserviceorganisations.Thusarathersmallnumberofresponsesfromdiverseorganisationswerereceived.Thesurveyresultswereasfollows:

Questions Yes No

AreyouoryourorganisationfamiliarwiththeconceptofLinkedOpenData(LOD)? 25 3

DoyouoryourorganisationknowofanyLODprojectsorinitiativesinyourcountryinthefieldofculturalheritage?

19 9

Haveyouor yourorganisationhadexperienceofusing LOD in connectionwith yourcollections?

6 22

Have you or your organisation had experience of publishing LOD in connectionwithyourcollections?

4 24

DoesyourorganisationplantopublishLODinthenearfuture? 21 7

DoesyourorganisationplantoconnectwithnewLODsources in thenear future? (1didnotanswerthisquestion)

14 13

In summary,most respondents to theAthenaPlus survey said that they (or their organisation) arefamiliarwithLinkedOpenDataandknewofrelatedprojectsandinitiativesintheircountry.Butonlyfew had first-hand experience with LOD. At the same time, most had plans to publish and/orconsumeLODinthenearfuture.

Sixteen respondents answered an open question on their expectations from connecting own datawith LOD resources. According to the survey authors the most common expectations related to“enlargingaccessibilityofdatainabroadercontext,increasingthevisibilityofcollections,extendthe

101AthenaPlus(EU,CIPBestPracticeNetwork,3/2013-8/2015),http://www.athenaplus.eu



semantic relations between various collections, development of cross-domain interdisciplinarynetworks of knowledge, possibility of re-contextualizing the resources for improved researchinfrastructure.Recognizedasanaddedvaluefortheowncollectionswasthepossibilitytoenrichowndata via (inter)national connections. One replymentioned the prospect of easy access to valuableinformationforscientificresearchandthepurposetocreateeducationalapps.”

Somerespondentsalsoconsideredpossibledisadvantages,whichincludedlossofcontroloverownpublisheddata, adecrease indataquality due to links tonon-qualified sources, or anoverloadoflinkswhichmightcausealossofvisibilityand/oraccessibility.

ARIADNEresultsforarchaeology

OneobserveroftheSemanticWebcommunitynotes:“Incontrasttotheculturalheritagesectorakamuseums, the Semantic Web has seen less uptake in archaeology. This could be becausearchaeologists tend to focus on analysis and recording of the data rather than dissemination.Experiences are mostly limited to spreadsheets, relational databases and/or spatial datamanagement.Manyacademicarchaeologistsremainprotectiveoftheirdataespeciallywhen ithasnot been published in traditional media. The complexity of combining siloed resources may beoverwhelming”(Solanki2009).

However,researchersarenotnecessarilytheprimarytargetgroupofLinkedDataawarenessraisingactions. The online survey reported in ARIADNE’s “First Report onUsers’Needs” (ARIADNE 2014a[April2014])hadonequestionabouthowhelpfulresearchersanddatamanagersperceivedifferentservices ARIADNE might provide. Among nine options there was “Improvements in linked data”,defined as “interlinking of information based on Linked Datamethods (i.e. methods of publishingstructureddatasothatitcanbeinterlinked)”.

Notsurprisingly,thisoptionwasatthebottomoftheresearchers’listofperceivedhelpfulness,onlythe service option “Content recommendations based on collaborative filtering, rating and similarmechanisms” fared worse. But of the over 470 researchers who answered the question still 37%thought“Improvements in linkeddata”couldbe“veryhelpful”and43%“ratherhelpful”(ARIADNE2014a:114).Thegoodresultsfor“Improvementsinlinkeddata”indicatethatinterlinkingofresearchresults is generally relevant to researchersand,arguably, thatquite some researchershadalreadyheardaboutLinkedDataasanovelwayofinterlinkinginformation.

An additional survey addressed repositorymanagers that are a considerably smaller target groupthan researchers. The survey received 52 sufficiently filled questionnaires, hence a good responsebut certainly not representative. The managers were asked if their repository and clients couldbenefit from services ARIDANE might provide, presenting the same list of service options as thesurvey of researchers. Among the managers who answered the question (32), the option“Improvementsinlinkeddata”faredbetter:itcameinonpositionfiveofthenineoptionswith39%“veryhelpful”and39%“ratherhelpful”.Thefavouritewas“ServicesforGeo-integrateddata”,52%“veryhelpful”,32%“ratherhelpful”(ARIADNE2014a:141).

The repositorymanagers in general weremore sceptical about potential improvements, but theyappreciated“Improvements in linkeddata” considerablymore than the researchers.Asnoted, theresultsforthedatamanagersarefarfromrepresentative.ButwethinkthattheyareindicativeandaddtoourviewthatdatamanagersareamorerelevanttargetgroupfortheLinkedDataapproachthan researchers. Data managers are active in different contexts, digital archives of the researchcommunity, repositories of individual institutions (e.g. university, research center), and largearchaeological projects in need of systematic and long-term data management. Within ARIADNE,consultancy and training for Linked Data has beenmainly given tomanagers of institutional data



resourceswith regard to vocabularies that are being used for themetadata of the resources, e.g.relatedtothemappingofthevocabulariestotheArt&ArchitectureThesaurus.

IntheARIADNEportalssurveyforthe“SecondReportonUsers’Needs”(ARIADNE2015a)23expertsof project partners (18 of which archaeologists) studied existing information portals, defined aswebsites that provide access to content ofmore than one institution or project. The aimwas toidentifygoodpracticesandgivefurtherideasforthedevelopmentoftheARIADNEdataportal.Someparticipants considered Linked Data for integrating information within the portal and linking toexternalresources.ThestatementsaddressedthepotentialoftheLinkedDataapproachaswellasthecurrentlackofawarenessofthebenefitsofsuchdata;alsotheneedofhigh-qualityLinkedDatawasmentioned(ARIADNE2015a:103-104).

The suggestions of the survey participants concerning Linked Data were summarised in threerecommendationsfortheARIADNEdataportalandevaluatedbyprojectpartners(28experts)withregardtotheirrelevanceandtime-horizon(ARIADNE2015e:282-287).Amongthetop-rankedofall34 recommendations of the portals survey was “Deploy Linked Open Data (LOD) to integrateinformationwithintheportalandtolinktoexternalresourceswhichfollowLODprinciples(e.g.HTTPURIsandRDF)”.79%oftheevaluatorsconsideredthisasrelevantand86%thoughtthatitmightbeachievedwithin the formal duration of the project (until January 2017). The evaluatorswere lessconfident with regard to encouraging a wider uptake of LOD principles among archaeologicalinstitutionsandprojects,butabout60%expectedthattheprojectwillpromotethis.

6.1.3 Briefsummaryandrecommendations

Briefsummary

LinkedDataenableinteroperabilityofdispersedandheterogeneousinformationresources,allowingthe resources to become better discoverable, accessible and re-useable. In the fragmented datalandscapeofarchaeologythis issubstantialvalueproposition.IntheARIADNEonlinesurveyontopof the expectations of the archaeological research community from a data portal were cross-searchingofdataarchiveswithinnovative,morepowerfulsearchmechanisms.ButsuchexpectationsarenotnecessarilyassociatedwithcapabilitiesofferedbyLinkedData.Thereforethegapbetweenadvantagesexpectedfromadvancedservicesand“buy in”andsupportof theresearchcommunityforLinkedDatamustbeclosedbytargetedactions.

A small survey of the AthenaPlus project (2013) indicated that cultural heritage organisations arealready aware of Linked Data, but few had first-hand experience with such data. Among theexpectationsfromconnectingownandexternalLinkedDataresourceswereincreasingthevisibilityof collections and creating relations with various other information resources. Some respondentsalso considered possible disadvantages, e.g. loss of control over own data or a decrease in dataqualityduetolinkstonon-qualifiedsources.

IntheARIADNEonlinesurvey(2013)“Improvements in linkeddata”, i.e. interlinkingof informationbasedonLinkedDatamethodstoenablebetter informationservices,wasconsideredmorehelpfulbyrepositorymanagersthanresearchers.Researchersofcourseperceiveinterlinkingofinformationas important, but may not see this as an area for own activity. Indeed, we think individualresearchersandresearchgroupsshouldnotbeaprimaryfocusofLinkedDatainitiatives.Managersofdigitalarchivesoftheresearchcommunityandinstitutionalrepositoriesaremuchmorerelevanttargetgroups.Furthermoredatamanagersoflargeandlong-termarchaeologicalprojectsshouldbeaddressedastheywillalsoconsiderrequiredstandardsfordatamanagementandinterlinkingmorethoroughly.



Recommendations

o AddressthehighlyfragmentedlandscapeofarchaeologicaldataandhighlightthatLinkedDatacanallowdispersedandheterogeneousdataresourcesbecomebetterintegratedandaccessible.

o Consider as primary target group of Linked Data initiatives not individual researchers butmanagersofdigitalarchivesandinstitutionalrepositories.

o Includealsodatamanagersand IT staffof largeand long-termarchaeologicalprojectsas theywillalsoconsiderrequiredstandardsfordatamanagementandinterlinkingmorethoroughly.

6.2 ClarifythebenefitsandcostsofLinkedData

OnetargetedactiontohelpclosethecurrentLinkedDataadoptiongapinthearchaeologicalsectorcouldberemovingthewidespreadnotionofanunfavourableratioofcostscomparedtobenefitsofemploying Semantic Web / Linked Data standards for information management, publication andintegration.WhilethestandardshavematuredandbecomemuchbetterapplicablethisnotionisstillprevalentandabarriertowideradoptionoftheLinkedDataapproach.

6.2.1 Thenotionofanunfavourablecost/benefitratio

Inapapertitled“IsParticipationintheSemanticWebTooDifficult?”,publishedin2002,theauthorsemphasised the need of lowering the entry barrier for cultural heritage organisations, especiallysmall ones, by offering significant added value and advantages over established ways of contentmanagement and publication (Haustein & Pleumann 2002). The authors note that initial stepstowardstheSemanticWebwillrequiresomeextraeffortand,therefore,“thesystemneedstoensurethatthiscostisoutweighedbythegainforthecontentprovider.Thisgainshouldnotcounttoomuchonthenetworkeffectof theSemanticWeb,becausethiseffectmighttakesometimetoreallypayoff.Instead,thegainhastobeimmediatelyvisibletothecontentprovider.”

IntheDigiCULTForumthematicissue“TowardsaSemanticWebforHeritageResources”(2003)theposition paper stressed that it is difficult to legitimate investment of institutions in the SemanticWeb, becauseover thenext five years itwouldbring little benefit (Ross 2003).ADigiCULT Forumassessmentin2004ofthereadinessofheritageinstitutionsforseverale-culturetechnologiesarguedthat SemanticWeb technologieswouldbeadoptedprimarilyby large institutions in a longer-termperspectiveof6ormoreyears(Geser2004).

WithregardtoanarchaeologicalsemanticWebJulianRichards in2006notedanincrease inonlineavailabledocumentsandarchivessothat“thereshouldbenoshortageofcontentwithwhichtobuildsuch aweb”; however “archaeology could get left behind if the rewards for creating themark-upnecessarytomaketheSemanticWebarealityareonlyevidentinthecommercialsector.ThesectoriscurrentlymorelikelytoparticipateinBerners-Lee’svisionthroughthecreationofsemanticmark-upfor informationaboutmonumentaccessarrangements,openinghoursandfacilities for thetourismindustrythanforacademicresearch”(Richards2006:977).

Reasons for thedoubtsofaquickadoptionof SemanticWeb standardsand technologies includedstillon-goingstandardizationwork,need for specialistknowledge, littleexperienceof implementa-tion under real world conditions and, in particular, expected high costs of conversion of legacymetadataandknowledgeorganizationsystemssuchasthesauritoSemanticWebstandards.



6.2.2 Lackofcost/benefitevaluation

Unfortunately, littleefforthasbeen invested so far tomakeclear cost /benefit ratiosofdifferentlevels andways inwhich LinkedData can be produced and employed. Among the exceptions is amodelthatconsiders“pay-offpoints”offiveescalatinglevelsatwhichinformationcanbeformalized(Isaksenetal.2010a/b).Thepurposeof themodel is toencouragea step-wiseadoptionofLinkedData principles, including for small-scale data sources (i.e. “small tail” data sets). The authorsconsiderthat“(at least)fiveescalating levelsofsemanticformalizationcanbe identified,eachwithdiffering requirements and benefits for the implementer: i. Literal Standardization, ii. InstanceURIgeneration, iii. Canonical URI mapping, iv. RDF generation, and iv. Database-schema-to-Ontologymapping”(Isaksenetal.2010a).

In this scheme(i)means thecreationanduseofa locallydefinedrestrictedvocabulary (e.g. listofterms or thesaurus), (ii) the creation of web-accessible unique identifiers for the proprietaryvocabularyterms,and(iii)mappingofthetermstoestablishedconcepts/termsofanacknowledgedauthority.ThesuggestedapproachseemsatoddswiththeLinkedDataprinciplethatprojectsshouldwherever possible re-use established vocabulary, however “normalization” of terms will often benecessarywhenattempting to integratedifferent legacydatasets. Thiswas thecase in theRomanPorts in theWesternMediterraneanProject (Isaksenetal.2009) towhich theauthors refer in thediscussionofthesuggestedschemeofsemanticformalization.

The authors emphasise “that Linked Data – hitherto seen as the simplest semantic approach – isrelativelyadvanced in thisscheme.Wearguethatdataprovidersshouldbeencouragedtomigratetowardsfullsemanticformalizationonlyastheirrequirementsdictate,ratherthanallatonce.Suchan approach acts as both a short and long-term investment in semantic approaches, in turnencouraging increased community engagement. We also propose that for such processes to beaccessibletodata-curatorswithlowtechnicalliteracy,assistivesoftwaremustbecreatedtofacilitatethesesteps”(Isaksenetal.2010a).

The authors also address benefits and costs (or, rather, requirements) of the different levels ofsemanticformalization,althoughonlygenerically.Forexample,thatRDFgenerationallowsmachinestoexploittheURIlinkagefordataaggregationanddiscovery,butrequiresabasicgraspofontologicalmodelling,selectionand/orcreationofpredicateURIs,toolsorscriptingfortheRDFgeneration,andmaybenew/unfamiliarRDFdatastoragemechanisms.

ThesuggestedapproachofastepwisemigrationtowardsLinkedDataseemsreasonable.Butwithoutamethodforevaluatingthe“pay-offs”intermsofthecost/benefitratio,andanumberofreferenceexamples, it will remain theoretical and of little help in driving “buy in” of potential Linked Dataproviders.

ThekeypointoftheapproachistolookfordifferentlevelsatwhichLinkedDatacanbeemployed.Inthis regard Eric Kansa of the archaeological data publication platform Open Context provides ahelpfuldiscussionofwhatcanbeconsideredasmediumandhigh-levelroutestoLinkedData(abovethelow-levelsemanticformalizationsmentionedbyIsaksenetal.).

Kansa(2014a)seesthemedium-levelrouteinannotationandcross-referencingofdatausingsharedcontrolled vocabularies,while the high-level is represented by employing the CIDOC CRM to aligndatasetsbasedonsharedconceptualmodelling(leveliv.“Database-schema-to-Ontologymapping”inthemodelsuggestedbyIsaksenetal.2010a).ReferringtoexperiencesfromOpenContextprojectsKansaisconvinced“thatvocabularyalignmentcanhelpresearchersmore,atleastinthenear-term,thanaligning datasets to elaborate semanticmodels (via CIDOC-CRM)”.At least it allows reaching



“some lower-hanging, easier to reach fruit in our efforts to make distributed data work bettertogether”and“meetmoreimmediateresearchneeds”.

One example of such a project employed annotations to common vocabularies to enable theintegration and comparison of zooarchaeological datasets from 17 sites (in total over 294,000records of bone specimens). Each dataset had its own organization (schema) and used somewhatdifferent proprietary vocabulary/terminology. The project annotated dataset-specific taxonomiccategories with Web URIs for animal taxa curated by the Encyclopedia of Life102, annotatedclassifications of bone elements with concepts of the Uber Anatomy Ontology (UBERON)103, andemployed a vocabulary developed by Open Context for bone fusion, sex determinations andstandard measurements. The vocabulary alignments provided the basis for data integration andcomparison across the different datasets (Arbuckleet al. 2014; Kansaet al. 2014;Whitcher-Kansa2015).

Concerning the CIDOC CRM, the high-level route of aligning datasets based on shared conceptualmodelling, despite its increasing adoption little is known about the cost / benefit ratio. Whileconsiderablebenefitshavebeenreportedinsomecases,thecostsideisusuallynotaddressed.

Forexample,Jordaletal.(2012)reportbenefitsandnewopportunitiesopenedupbytheCRM-basedintegrationofethnographiccollectionsheldbytheMuseumofCulturalHistory inOslo.ConnectingthecollectionsviaaCRM-basedmodelallowsthecuratorsintegratedaccesstothelegacycataloguesand databases, and the model also guides the registration of new items. The integration of thecollectionsalso“givesabetterbasisfortellingastoryforeachartefact”,and“providesapossibilitytodoresearchontheobjectswithascomplete,accurateandrichdataaspossible”.

Other institutions have achieved a lot by applying the CIDOC CRM to integrate large andheterogeneous datasets, enable advanced search on their website, and participate in culturalheritagewebportals.OneoutstandingexampleinthisregardisArachne,thecentralobjectdatabaseof the German Archaeological Institute (DAI) and the Archaeological Institute of the University ofCologne104.TheCIDOCCRMbasedinternalintegrationofdataallowsadvancedexplorationofamassofheterogeneousinformationresources.ArachnealsoparticipatesinCLAROS-ClassicalArtResearchOnlineServices(launched inMay2011)105whichprovidesaportal forsearchingseveralsourcesforClassicalstudiesbasedontheLinkedDataapproachandCIDOCCRM.

Oldman&Rahtz(2014)highlightthattheCLAROSproject“establishedthecredentialsoftheCIDOCCRM standard as a semantic framework that can harmonise data frommany different institutionswhileprovidingaricherenvironment(whencomparedtoitsdigitalsources)inwhichtoexploreandresearch cultural heritage data”. But the CLAROS Linked Data based search environment offersratherlimitedresearchfunctionality.TheResearchSpaceproject106,inwhichDominicOldmanservesasprincipalinvestigator,aimstoenableadvancedexplorationandresearchofCIDOCCRMmediatedculturalheritagedata.

102EncyclopediaofLife,http://eol.org103UBERON-UberAnatomyOntology,http://uberon.org104Arachne,http://arachne.uni-koeln.de105CLAROS,http://www.clarosnet.org;http://data.clarosnet.org106ResearchSpace,http://www.researchspace.org



6.2.3 Collectingexamplesofbenefitsandcosts

BenefitsofLinkedData

The basic assumption of Linked Data is that the usefulness and value of data increases themorereadily it can combinedwith relevant other data. The Linked Data approach of using stable URIs,typed RDF links and common vocabulary greatly supports benefits from bringing together relatedinformation.Berners-LeedescribedbenefitsofLinkedDatawithphrasessuchas“toprovidecontext”orthatusers“candiscovermorethings”(Berners-Lee2006andadditionon5-stardatain2010).

Indeed,convincingtangiblebenefitsofLinkedDatamaterialiseifinformationproviderscandrawonown and external data for enriching services. A prominent early example is that the BBC usedDBpedia(WikipediaLinkedData)107undMusicBrainzLinkedData108toenrichtheinformationoftheirmusicpages(Kobilarovetal.2009;Raimondetal.2013reportonBBC’suseofLinkedDataforotherservices). An example from themuseumworld is the Smithsonian American ArtMuseum (SAAM)thatenriches theirartistpageswith identifiersof theGettyUnionListofArtistNames (ULAN)andinformationfromDBpediaandNewYorkTimesLinkedData(Szekelyetal.2013;Zaino2013).

Szekely et al. (2013) summarize the benefits for the SAAM as follows: “the linked data providesaccess to information that was not previously available. The Museum currently has 1,123 artistbiographiesthatitmakesavailableonitswebsite;throughthelinkeddata,weidentified2,807linkstopeople records inDBpedia,whichSAAMpersonnelverified.TheSmithsoniancannow link to thecorrespondingWikipediabiographies, increasing thebiographies theyofferby60%.Via the links toDBpedia, they now have links to the New York Times, which includes obituaries, exhibition andpublicationreviews,auctionresults,andmore.Theycanembedthisadditionalrichinformationintotheir records, including 1,759 Getty ULAN identifiers, to benefit their scholarly and publicconstituents.”

ThissuggeststhatthebenefitofLinkedDatamaysomehowbecalculatedbasedonthe increase inrichnessof information servicesperdatasetadded,alsoconsideringdifferentbeneficiaries suchas(inthisexample)arthistorians,journalistsandpeoplegenerallyinterestedtolearnaboutartistsandartworks.

Similar examples should be collected or developed as Linked Data use cases for datasets ofarchaeological research projects and archives/collections. It seems clear that popular Linked DataresourceslikeWikipediamaynotbeappropriateforpurposesofarchaeologicalresearch.Butthereareotherresources,forexample,amongtheextensiveLinkedDataofthebio-scienceswhichmightbe exploited for relevant research use cases concerning human, animal or plant remains (e.g. theexampleofzooarchaeologicalLinkedDatareportedinKansaetal.2014).

ButsomedifferencesbetweenbenefitsofenrichingviaLinkedDatamuseumorarchiveinformationandintegratingresearchdatashouldbenoted.Culturalheritageinstitutionscanbenefitfrommakingtheir collections more meaningful and relevant to end-users by adding external contextualinformation(linkstorelatedcontent). Inawebofrichly interlinkedinformationthein-cominglinkscan also leverage usage of own content. This is fully in line with the institutions’ mission tocommunicatecontextualisedculturalheritagetoanaswideaspossibleaudience.

In the realm of research the benefits of Linked Data should be reflected in terms of researchdividends that can be gained by interlinking data. Such dividends for example are discovery of

107DBpedia,http://wiki.dbpedia.org108LinkedBrainz-MusicBrainzinRDFandSPARQLhttp://linkedbrainz.org



relationsbetweenresearchdataworthexploringfurther,combinationofdatafromdifferentprojectsin ways that enable interesting new lines of research, different views on data from variousdisciplinary perspectives suggesting interdisciplinary approaches, etc. (see the discussion of searchvs.researchinSection6.6).

CostsofLinkedData

InordertoevaluatethecostsofLinkedDataproviders,informationaboutthedifferentcostfactorsanddriversshouldbecollected.AgoodunderstandingofthecostsofdifferentLinkedDataprojectsmay help to possibly reduce the costs, for example, by providing dedicated tools, guidance andsupportforcertaintask.

Thecostsingeneralconcerntheacquisitionoftheexpertiseandtheworkeffortandtoolsrequiredfortheactualgeneration,publicationand interlinkingofthedata.Basicsteps intheprocessaretoselectrelevantdata,cleanit,designtheURIs,convertthedatatoRDF,storeandmakeitaccessible,mapproprietarytermstoestablisheddomainvocabulary,andfindandcreatelinkstorelateddataontheWeb109(seeSection3.5).

Fortheprocessstepsinformationaboutthecostsshouldbecollectedandanalysed,takingaccountofprojectsofdifferenttypesandsizes.Asanexampleofrequiredinformation:IntheMultimediaNE-Culture project several legacy datasets from different institutions have been converted to LinkedData and integrated (Omelayenko 2008): It was found that nearly every dataset required somedataset-specificcodetobewritten.Butbyidentifyingandseparatingconversionrulesthatcouldbere-used the overall effort was reduced considerably. Nevertheless, it has been estimated that askillful professional who uses a state-of-the-art conversion support tool (in this case, AnnoCultor)needed around four weeks to transform a major museum database, creating for this purpose adedicatedconverterof50-100conversionrulesplussomecustomcode.

Somenewmethodsandtoolshavereducedconsiderablythecostsofdataconversion,publication,annotation and linking. For example, Van Hooland et al. (2012a) of the Free Your Metadatainitiative110 argue that the interactive data cleaning and transformation tool OpenRefine111 “hasmade data cleaning and reconciliation available for the masses”. Clearly data cleaning, trans-formationandreconciliation(matchingentitieswithotherLinkedData)areessentialstepsinLinkedDatageneration.TheauthorsillustratethecasewithmetadataoftheCooper-HewittNationalDesignMuseum,NewYorkandthePowerhouseMuseum,Sydney(VanHoolandetal.2012aand2012b).

Numerous other tools are available ranging from tools for specific tasks to comprehensive LinkedData generation, management and publication platforms. The proliferation of tools means thatpotential Linked Data providers need expert advice on what to use (and how to use it) for theirpurposesandspecificdatasets,takingaccountalsoofexistinglegacysystems,standardsinuse,etc.

ParticularlyrelevantinthiscontextareapproachesthatallowexploitinglegacydatabasesandavoidkeepingandmanagingRDFdataseparately inadedicateddatabase(triplestore).VarioussolutionsareavailabletooutputdatainRDFfromexistingdatabases(Sahooetal.2009;Micheletal.2013)112.This requires amapping of the database to RDF, whichmay be created automatically (for simpledatabases)butmoreoftenneedsanexpertmappingtoadomainontologyinRDFSchemaorOWL.109W3C(2014)WorkingGroupNote:BestPracticesforPublishingLinkedData,9January2014,

https://www.w3.org/TR/ld-bp/110FreeYourMetadata,http://freeyourmetadata.org111OpenRefine,http://openrefine.org112OneexampleisD2RQ-AccessingRelationalDatabasesasVirtualRDFGraphs,http://d2rq.org



Asanexampleofanarcheologicaldatabase,theLaboratoireArchéologieetTerritoires,UniversitédeTours-CNRS,FranceaimstoopenuptheirArSol-ArchivesduSol(SoilArchives)system113basedonamappingofconceptsoftherelationaldatabasetotheCIDOCCRM.ThismappingisbeingusedtoquerythedatabaseemployingSPARQL-to-SQLrewrites(LeGoffE.etal2015;Marletetal.2016).Theapproachavoidstheextract-transform-load(ETL)processforexportingdatainanRDFstoreandforupdating it when data changes. The researchers employ the Ontop114 platform developed by theKnowledge Representation meets Databases (KRDB) research group at the University of Bozen-Bolzano(Bagosietal.2014).ThesameapproachandplatformisbeingusedbytheEPNetproject115(Calvaneseetal.2015;Calvaneseetal.2016).

Effective and easy-to-use tools are of utmost importance for reducing the costs of core tasks ofLinkedDatageneration,publicationandlinking.ButadviceonhowtobestapproachothertaskssuchasURIdesignorvocabularyselectioniscriticalaswell.

HereisnottheplacetoaddressallstepsinthesocalledlifecycleofLinkedDatafromdataselectiontoRDFpublicationanduse,particularlybecausecostfiguresarehardtocomeby.Asanexample,astudy by PricewaterhouseCoopers for the Interoperability Solutions for European PublicAdministrationsprogrammelooked intobusinessmodelsfor linkedopengovernmentdataservices(Archer et al. 2013). One of their research questions therefore concerned the costs of the LinkedDataservices,includingdevelopment,maintenanceandpromotion.

Thestudyinvestigated14casesbutdidnotbringoutthecoststructureoftheLinkedDataactivitiesbecausemost respondents did not separately account for this. Only the German National Librarygave figures for specific development tasks and on-goingwork for Linked Data provision116: Initialdevelopment including mappings between internal database format and RDF vocabularies,implementation of data conversions, and standards relatedwork consumed 221 person days; theestimatedeffortformaintenancewas1FTE(full-timeequivalent)butforthebibliographicserviceswhichincludedthesupplyofLinkedData;thecostspecificallyforthelatterremainedunclear(Archeretal.2014:3,30and58).

Afinalimportantpoint,thediscussiononcostsofLinkedDataingeneral(includingabove)centresonthedataandvocabularyproviders.But intheLinkedDataecologyalsothecostsofpotentialusersneed tobeconsidered.Asone respondent toadiscussiononwhydataproviders shouldcarry thecostsofpublishingLinkedDataemphasised,“inthecurrentstateoftheworld, itcomeswithaddedcostsfortheconsumersaswell.Mostdevelopersdon’tknowmuchaboutRDFandsurroundingtoolsandstandards,sotheyhavetolearnaboutitinordertoconsumeyourdataset.Thesecostscaneasilyoutweighpotentialbenefits.Ofcourse, themissionof the linkeddatacommunity is tochange thatfactbypopularizingRDFtechnologiesandstandards,sothatmightnotbetrueanymore5yearsfromnow” (Samwald 2010). Another respondent seconded this by adding, “I don’tmean to say LinkedDataisnotthewayforward,Ijustdon’tthinkit’syetarepresentationthatlargenumbersofpeoplewould feel comfortable or capable of working with, given what they currently know, what theycurrentlydo,andtheyculturallycurrentlydoit…”(Hirst2010).

113ArSol-ArchivesduSol(SoilArchives),http://arsol.univ-tours.fr114Ontop,http://ontop.inf.unibz.it115EPNet-ProductionandDistributionofFoodduringtheRomanEmpire:EconomicandPoliticalDynamics

(ERCAdvancedGrantproject,3/2014-2/2019),http://www.roman-ep.net116LinkedDataServiceoftheGermanNationalLibrary,http://dnb.de/EN/lds



Costsofknowledgeorganizationsystems

Knowledgeorganizationsystems(KOSs),includingformssuchasthesauri(terminology),taxonomies(classificationsystems)andontologies(conceptualreferencemodels)playakeyroleinLinkedData.IndeedwithoutthesemanticsofKOSsawebofmeaningfulLinkedDatacannotbebuilt.ThereforeitisastonishingthatlittleisknownaboutthecostsofemployingKOSs.

As an example, in a special issue of the Bulletin of the Association for Information Science andTechnologypublished2014(ASIS&T2014)ontheeconomicsofKOSsnoneofthefivearticlesgivesanexample of the actual or estimated costs of a KOS. However, Denise Bedford in this bulletinelaborates in detail the assets and liabilities different types of “taxonomies” (her term for KOSs)generate,forexampleaflat listoftermsvs.athesaurus.Bedfordalsogivesanoverviewofgeneralcategories of costs involved, but states: “The actual costs of any taxonomy project are tied to itsorganizationalcontextandthescopeandscaleoftheeffort.Itisnotpossibleoradvisabletosaythatatypicalthesaurusprojectcanbecompletedfor$100,000orfor$500,000becausethereisno‘typicalthesaurus’”(Bedford2014:20).

Lackof solid knowledgeabout the costsof employingKOSshas a long “tradition” in the SemanticWeb (LinkedData) community.Forexample,TimBerners-Lee,WendyHallandNigelShadbolt, keyfiguresofthecommunity,intheirpaper“TheSemanticWebRevisited”(Shadboltetal.2006)addresstheissueofcostsbutcanonlygive“naïvebutreasonableassumptions”.Theyconsiderthatinsomeapplication“thecosts–nomatterhowlarge–willbeeasytorecoup.Forexample,anontologywillbe a powerful and essential tool in well-structured areas such as scientific applications. In certaincommercial applications, the potential profit and productivity gain from using well-structured andcoordinatedvocabularyspecificationswilloutweighthesunkcostsofdevelopinganontologyandthemarginalcostsofmaintenance.Infact,giventheWeb’sfractalnature,thosecostsmightdecreaseasanontology’suserbaseincreases. Ifweassumethatontologybuildingcostsarespreadacrossusercommunities,thenumberofontologyengineersrequiredincreasesasthelogoftheusercommunity’ssize.Theamountofbuildingtimeincreasesasthesquareofthenumberofengineers.Thesearenaïvebutreasonableassumptionsforabasicmodel.Theconsequenceisthattheeffortinvolvedperuserinbuilding ontologies for large communities gets very small very quickly”.They go on discussing thedifferencebetweendeepandshallowontologies,requiring“considerableeffort”(fortheontologicalconceptualization)and(unspecified)“effortbutovermuchsimplersetsoftermsandrelations”inthecaseofshallowontologies(Shadboltetal.2006:99).

Hepp (2007) addresses economic and other issues that constrain the development, adoption andmaintenance of useful ontologies and other KOSs. He notes that KOSs are regarded as centralbuildingblocksoftheSemanticWeb,andmuchhasbeenwrittenaboutthebenefitsofusingthem,but that there are substantial disincentives for building and adopting relevant KOSs. He discussesinterestinggeneralassumptions,butalsodoesnotgiveasinglecostfigure.

HeppassumesthatKOSsexhibitpositivenetworkeffects,hencetheirperceivedutilitywill increasewith the number of users. But convincing people to invest effort into building or using them isdifficult in the initial phase in which there is no or only a small user base. The utility for earlyadoptersislow,whereasadoptionmayrequireahighereffortthaninalaterphaseofdiffusionwhenpracticalusecasesandexpertiseareavailable.AtthatpointaKOSmayalsobemoreelaboratedandcover better the intended domain of knowledge. Particularly interesting are Hepp’s empiricallyconfirmed assumptions concerning the relation between the expressiveness of a vocabulary(ontology)andthesizeofthecommunitythatwilladoptit.



Basically,themoreexpressivetheontology,thesmallertheusercommunitywillbe,becauseoftheeffortnecessarytocomprehendandapplyit(arguablytheCIDOCCRMissuchacaseasdiscussedinSection6.3.3).Inpracticethiscomesdowntothefactthat“usefulontologiesmustbesmallenoughto have reasonable familiarization and commitment costs and big enough to provide substantialaddedvalueforusingthem”(Hepp2007:94),wherebigenoughmeansbothsufficientcoverageoftheintendeddomainandtheexistinguserbase.ArguablythisiswhysmallvocabulariessuchasFOAFandDublinCore(dcterms)aremostwidelyusedinsetsofLinkedData(Schmachtenberg2014a;seealsoCoyle2013ontheuseofDublinCoreinLOD).

ExcellentworkonthecostsofcreatingKOSshasbeendonebytheONTOCOMproject117.But theirhighlyelaboratedmodelofcostfactorsanddriversdoesnotincludethecostofactuallyemployingaKOSforpurposessuchasdatatransformationandlinking(cf.Simperletal.2012).


Briefsummary

There isawidespreadnotionofanunfavourable ratioofcostscompared tobenefitsofemployingSemanticWeb / Linked Data standards for informationmanagement, publication and integration.This notion should be removed as it is a strong barrier to a wider adoption of the Linked Dataapproach.

The basic assumption of Linked Data is that the usefulness and value of data increases themorereadily it can combined with relevant other data. Convincing tangible benefits of Linked Datamaterialiseifinformationproviderscandrawonownandexternaldataforenrichingservices.Thereareexamplesforsuchbenefits,e.g.inthemuseumcontext,butnotyetforarchaeologicalresearchdata. Importantly, intherealmofresearchbenefitsofLinkedDataare lessaboutenhancedsearchservices but research dividends, e.g. discovery of interesting relations or contradictions betweendata.

LinkedDataprojectstypicallymentionsomebenefits(e.g.integrationofheterogeneouscollections,enrichedinformationservices),butverylittleisknownaboutthecostsofdifferentprojects.Thereisaclearneedtodocumentanumberofreferenceexamples,forexample,whatdoesitcosttoconnectdatasetsviasharedvocabulariesor integratedatabasesthroughmappingthemtoCIDOCCRM,andhowdoesthatcomparetoperceivedbenefits?AlthoughvocabulariesplayakeyroleinLinkedDataastonishinglittleisalsoknownaboutthecostsofemployingvariousKOSs.

Somemethodsandtoolsappear tohavereducedthecostofLinkedDatagenerationconsiderably,OpenRefineormethodstooutputdatainRDFfromrelationaldatabases,forinstance.AsthereisaproliferationoftoolspotentialLinkedDataprovidersneedexpertadviceonwhattouse(andhowtouse it) for their purposes and specific datasets, taking account alsoof existing legacy systems andstandardsinuse.

Recommendations

o Proponents of the Linked Data approach should address the widespread notion of anunfavourable ratio of costs compared to benefits of employing Semantic Web / Linked Datastandards.

117OntologyCostEstimationwithONTOCOM,http://ontocom.sti-innsbruck.at



o Major benefits of Linked Data can be gained from integration of heterogeneous collections/databasesandenhancedservicesthroughcombiningownandexternaldata.Butexamplesthatclearlydemonstratesuchbenefitsforarchaeologicaldataareneeded.

o Inordertoevaluatethecosts,informationaboutthecostfactorsanddriversshouldbecollectedandanalysed.AgoodunderstandingofthecostsofdifferentLinkedDataprojectswillhelpreducethecosts,forexamplebyprovidingdedicatedtools,guidanceandsupportforcertaintasks.

o MoreinformationwouldbewelcomeonhowspecificmethodsandtoolshaveallowedinstitutionsreducingthecostsofLinkedDatainprojectsofdifferenttypesandsizes.

o Generalrequirementsforprogressaremoredomain-specificguidanceandreferenceexamplesofgoodpractice.

6.3 Enablenon-ITexpertsuseLinkedDatatools

There are already several showcase examples of Linked Data application in the field of culturalheritage(e.g.museumcollections)which,however,dependedheavilyonthesupportofexpertswhoarefamiliarwiththeLinkedDatamethodsandrequiredtools.AmuchwideruptakeofLinkedDatawill require approaches that allownon-IT expertsdomostof theworkwitheasy touse tools andlittle training effort. A number of projects have reported advances in this directionbasedondatamapping recipes, supportive tools and guidance material. Further progress may be achieved byintegratingLinkedDatavocabulariesintoolsfordatarecordinginthefieldandlaboratory.

6.3.1 LinkedDatatools:therearemanyandmostarenotuseable

LinkedDatatoolsisafieldofsoftwaredevelopmentthatislargelydominatedbyacademicresearchgroups and individual developers (e.g. in the context of a PhD thesis).While produced under theopen sourcebanner, theirwork rarely leads tomature,maintained and serviced tools or services.There is a lot of obviously immature and abandoned softwareof suchdevelopers onopen sourcesoftwareplatforms(e.g.GitHub,SourceForgeandothers)orprojectwebsites.Oftentheaimseemsnottobeaworkingsolutionbutanumberofpublicationsaroundthetoolorservicedevelopment.As Hafer & Kirkpatrick (2009) note, “Academic computer science has an odd relationship withsoftware: Publishing papers about software is considered a distinctly stronger contribution thanpublishingthesoftware”.Thehigheracademicrecognitionofpublicationsimpactsnegativelyonthecurationandlong-termavailabilityofsoftwarethatisproducedinthiscontext(Todorov2012).

Some academic open source projects are successful because they find a community of dedicateddevelopers or are developed further by a commercial spin-off, but relevant others would needinstitutional support and curation toensure sustainability (Katzetal. 2014;Wilson2014). In somerespects the development of semantic tools presents a quasi-Darwinian pattern of survival of thefittest.ThefieldofsemanticWikismayserveasarepresentativecase:AsectionofSemanticweb.orglists37semanticWikiprojects118ofwhich30(80%)appeartobedefunctorareinactivesincelong.Suchlistsareveryhelpfulbecauseseldomsoftwareprojectwebsitesindicatethatworkonatoolhasbeendiscontinuedormaybesupersededbyanotherproject,onanewwebsiteandrenamedtool.Inmostcasesofstillavailablesoftwareitremainsunclearifthetoolhasbeencompletedandisusable,orisanunstableprototypewithlimitedfunctionality,bugs,etc.

118Semanticweb.org:SemanticWikiprojects,http://semanticweb.org/wiki/Semantic_wiki_projects



TheLODAroundtheClock(LATC)projectwarnsthatalotofopensourceLinkedDatasoftwaretoolsarenot completed,well-testedandstable.Thedevelopersoften lose interest inaproject“leavingusersstrandedwithout improvementsorsupport” (LATC2012:10-11, includesa listofquestionstoconsiderintheevaluationofrelevanttools).LATC,LOD2119andotherprojectspresentselectedtoolsfordifferentphasesoftheLinkedDatalifecycle,buttheselectionisofteninformedbywhatprojectparticipantshaveonstock.Moreovertoolssuggestedbyprojectscompletedtwoorthreeyearsagomayalreadybesupercededbynewoneswithfeaturesthatareimprovedinsomerespects.

Inshort,newentries intherealmofLinkedDatashould lookwhichtoolsarebeingusedbysimilarotherprojectsandconsultwithexpertsinthefieldwhichoneswillfitbestfortheirdataandgoals.

6.3.2 Needofexpertsupport

ArguablyallLinkedDatashowcases in the fieldofculturalheritageso fardependedheavilyonthesupport of expertswho are familiarwith the requiredmethods and tools, often their own.Manyprojects have been by experts together with museums, starting with the path-breaking FinnishMuseums on the SemanticWeb project (Hyvönen et al. 2002) up tomore recent projects at theAmsterdamMuseum(deBoeretal.2012and2013),GothenburgCityMuseum(Damova&Dannells2011), Peter theGreatMuseumofAnthropology andEthnography in St Petersburg (Ivanov2011),RussianMuseum in St. Petersburg (Mouromtsev et al. 2015), Smithsonian American ArtMuseum(Szekelyetal.2013),naturalhistorymuseums in theNaturalEuropeproject (Skevakisetal.2013),and others.120 One reason for the strong presence of museums is that they wish to make theircollections more accessible to the public, and may more easily do this by drawing on popularresourcessuchasWikipediaviaDBpediaLinkedData.

AmuchwidergenerationanduseofculturalheritageandarchaeologyLinkedData,especiallyalsoforresearchpurposes, requiresappraochsthatallownon-expertstodotheworkwitheasytousetools and little training effort. But this may remain an illusory goal. As Eric Morgan, the leadresearcheroftheLinkedArchivalMetadata(LiAM)notes:"Linkeddatamightbea 'goodthing',butpeople are going to need to learn how toworkmore directlywith it" (Morgan 2014).He suggestspractical tutorials,hands-on trainingonhowLinkedDatacanbeput intopractice,andhackathonsinvolvingpractitionersandLinkedDataspecialists.

In short, turning substantial legacy collectionsor researchdatasets into LinkedData resourceswillhardly be possible without support of specialists, at least for some steps in the process. As asummary of a discussion on skills required for LinkedData puts it, “Realistically, formany people,expertiseneedstobebroughtin.Mostorganisationsdonothaveresourcestocallupon.Oftenthisisgoingtobecheaperthanup-skilling–asteeplearningcurvecantakeweeksormonthstonegotiatewhereassomeoneexpertinthisdomaincoulddotheworkinjustafewdays”(Stevenson2011).

6.3.3 ThecaseofCIDOCCRM:fromdifficulttodoable

Aspecialcaseofadifficultadoptionprocess is theCIDOCConceptualReferenceModel,which isacore for cultural heritage information exchange and integration. The CIDOC CRM is an ontologyrepresentedinRDFSchema(RDFS)andconsideredasakeyintegratorofheterogeneousdatasetsin

119LOD2-CreatingKnowledgeoutofInterlinkedData(EU,FP7-ICT,2010-2014),http://lod2.eu120SomeotherexamplesarelistedontheMuseumsandtheMachine-processableWebwiki,e.g.Auckland

Museum(NewZealand);BritishMuseum(UK),HarvardArtMuseums(USA);NationalMaritimeMuseum(UK)andothers,http://museum-api.pbworks.com/w/page/21933420/Museum%C2%A0APIs



theemergingwebofculturalheritageLinkedData.TheontologybecameanofficialISOstandardin2006 (ISO 21127:2006, updated in 2014), which is but one factor that contributed to its wideradoptionintheculturalheritagesector,includingarchaeology.

TheincreasinguseoftheCIDOCCRMinrecentculturalheritageLinkedDataprojectsisnoteworthy.InitsearlydaystheCIDOCCRMwasperceivedasdifficulttoapplybyresearchersandpractitionerswhowerenot involvedin itsdevelopmentandrelateddemonstrationprojects.Forexample, intheSCULPTEUR project (2002-2005) museum databases were mapped to the CRM to implementconcepts-based cross-collections search & retrieval. The implementers reported that “mapping iscomplex and time consuming. The CRM has a steep learning curve, and performing the mappingrequiresagoodunderstandingofbothontologicalmodellingaswellasthesourcemetadatasystem.Eventually the assistance of a CRM expert was required to complete and validate the mappings”(Sinclairetal.2005).

Indeed,theCIDOCCRMisacomplexontologythatrequiresagoodunderstandingofitsevent-centricmodellingapproachaswell ashow toapply,extendor specialise theontology foraparticularusecase,ifrequired.ResearchersoftheBRICKSproject(2004-2007)notedtheabstractnessoftheCRMconcepts and lack of technical specification as factors that could impede the goal of enablinginteroperability across heterogeneous databases (Nußbaumer & Haslhofer 2007; see alsoNußbaumeretal.2010).

Similarstatementscanbefoundelsewhere,forexample,onerespondenttoLeifIsaksen’ssurveyoncultural heritage and archaeology Semantic Web projects wrote: “CIDOC CRM is bloody hard tounderstandandusewithzerotoolsupportavailableatthetime.Museumbodsareunderstandablynot knowledge engineers, so require lots of support” (in Isaksen 2011: 203). On the other hand,DominicOldman (2012) notes that someof the issues pertain to “a lack of domain knowledge bythosecreatingculturalheritagewebapplications.TheCRMexposesarealissueintheproductionandpublicationofculturalheritageinformationabouttheextenttowhichdomainexpertsareinvolvedindigital publication and, as a result, its quality (…) The CRM requires real cross disciplinarycollaborationtoimplementproperly–andthistypeofcollaborationisdifficult.”

MeanwhileanumberofexemplaryCIDOCCRMusecases,availabledocumentationandsharingofknow-howamongpractitionershaveenabledmoreprojects largeandsmallapplying theontology.Howevernewcomerswillstilloftenneedexpertguidance,ashasbeengiventoARIADNEpartnersbyFORTH-ICS’CentreforCulturalInformaticsonmodelingscientificarchaeologicaldata121.

6.3.4 Progressthroughdatamappingtoolsandtemplates

Projectsondatabasesofheritage collections reported considerabledifficulties ingetting to LinkedDataandarchaeologicalresearchdatasetsarguablyposeevengreaterchallenges.Forexample,thedatasetsthatweremappedintheRomanPortsintheWesternMediterraneanProjectaredescribedas follows: “While the datasets all pertain to the same domain, they frequently employ mixedtaxonomies and are heterogeneously structured. Normalization is rare, uncertainty frequent andvariant spellings common. Different recording methodologies have also given rise to alternativequantificationanddatingstrategies.Inotherwords,itisatypicalreal-worldmixed-contextsituation”(Isaksenetal.2009).

121Cf.ARIADNE(2014b),website:Modelingscientificdata:workshopreport,12September2014,

http://www.ariadne-infrastructure.eu/News/Modeling-scientific-data



But a number of projects have reported advances toward the goal of enabling non-experts applysemantic standards and tools. The datamapping tools thatwere developed and employed in theRomanPortsproject“haveprovenremarkablysuccessfulagainstabroadrangeofsampledatasetsfromfourdifferentcountries(UK,Spain,France,Italy).Themostimportantachievementhasbeentoenable domain experts to provide data derived in different contexts as ontology-compliant LinkedData extremely quickly and sustainably. Previous attempts to produce homogeneous RDF havegenerallyrequiredalengthyandexpensivemappingprocessagainstoneortwolargeresources.Wefeelthatmakingitpossiblefor‘thelongtail’ofarchaeologicaldataisavitaltaskintheLinkedDatarevolution”(Isaksenetal.2009).

Similarly, the Linked Data toolkit developed in the STELLAR122 project has been reported to allownon-expertusersmappingandextractingarchaeologicaldatasetstoXML/RDFconformingtoCIDOCCRM, CRM-EH (English Heritage) or CLAROS CRM Objects concepts and relations. The toolkitcomprisesofanopensourcesoftwaretool(StellarConsole)andasetofcustomizabletemplates.Theapproach takenwas to identify a set of commonly occurring patterns in domain datasets and theCIDOCCRM,andexpresstheminasetofmappingtemplates.

Tudhopeetal.(2013)notethatwiththeCIDOCCRMthesamesemanticsunderlyingculturalheritagedatasets can bemapped in different ways, which raises barriers for semantic interoperability theCRMaimstoenable.CRMadoptersneededmappingguidelinesandtemplatesforgeneralusecasesintheirdomain(e.g.archaeology).ThereforetheSTELLARprojectmadeavailableafacilityforuser-definedtemplatesaswellashelpfultutorialswithworkedexamples123(Bindingetal.2015presentindetailthetemplateuseforarchaeologicaldatasetsandacasestudywithnonexpertusers).

TheSTELLARtemplateshavebeenadaptedandusedbyotherprojects.Forexample,theArcheoInfproject124 aimed todevelopadatabase that combines and integrates, throughmappings toCIDOCCRM,dataofarchaeological surveysandexcavationsconductedbyGermanuniversity institutesofclassical archaeology. Adapted STELLAR templates allowed exporting datasets tagged with CIDOCCRMmappings in XML/RDF (Carver 2013; Carver& Lang 2013).Other projects that employed theSTELLAR toolkit for LinkedData generationwere Colonisation of Britain (digitisation and semanticenhancement of a major research archive)125 and the SKOSification of the thesaurus used withZENON,theonlinepublicaccesscatalogoftheGermanArchaeologicalInstitute(Romanello2012).

6.3.5 Needtointegratesharedvocabulariesintodatarecordingtools

WewillalsoneedtoseemoreprogresswithregardtointegratingLinkedDatavocabularies indatarecordingtools.It iswidelyheldthatarchaeologistsexhibitanaversiontouseunfamiliarsemanticsandprefertodeveloptheirownvocabulary.Theargumenttypicallyisthatthisisnecessarybecauseof their specific research questions. Frederick W. Limp even thinks that “the reward structure inarchaeological scholarship provides a powerful disincentive for participation in the development ofsemantic interoperability and, instead, privileges the individual to develop and defend individualterms/structuresandcategories”(Limp2011:278).

122STELLAR-SemanticTechnologiesEnhancingLinksandLinkedDataforArchaeologicalResourcesproject(UK,

AHRC-fundedproject,2010-2011),http://hypermedia.research.southwales.ac.uk/kos/stellar/123HypermediaResearchUnit,UniversityofSouthWales:STELLARApplications,

http://hypermedia.research.southwales.ac.uk/resources/STELLAR-applications/124ArcheoInfproject,http://www.ub.tu-dortmund.de/archeoinf/125Archaeogeomancy.net(2014):ColonisationofBritain,30May2014,

http://www.archaeogeomancy.net/2014/05/colonisation-of-britain/



The reticence to use vocabularies that are based on semantic standards is augmented by aperceptionthat thiscanbedifficult, timeconsumingandhaveno immediatepracticalbenefit.TheteamofOpenContext in thedevelopment theirarchaeologicaldatapublicationplatformcollectedviews and practical experiences of many archaeologists, cultural resource managementprofessionals, museum curators and others. The results across all participants suggested “littlemotivationor interest inhavingresearchers ‘markup’theirowndatatoalignthesedatawithmoregeneralWeborsemanticstandards”.Ratherprojectparticipants“generallysawthisasasomewhatabstract goal, disconnected from their immediate needs, and usually felt such semantic andstandardsalignmentstoodtoofaroutsideoftheirareaofexpertise”(Kansa&Whitcher-Kansa2011:5-6).

The Federated Archaeological Information Management Systems project (FAIMS, Australia) inworkshops with potential users found that archaeologists would appreciate tools that allow highflexibilityandcustomizationtoaccommodatetheirestablishedresearchpractices.Littleenthusiasmwasperceivedforadoptingcommondatastandardsandterminology,e.g.torecordanagreedsetofattributesaboutexcavationcontextsorartefacts(Rossetal.2013:111-114).

The resultsmade the FAIMS team rethink their approach to semantic interoperability, whichwasinitiallyplanned tobuildarounda stable (ifextensible) coreofdata standards,data schemataanduser interfaces.Toaccommodateboth flexibilityand interoperability,FAIMSmobiledatarecordingsoftware nowprovides sophisticated tools tomap data to shared vocabularies as it is created. Astheydescribethetools,“UsinganapproachborrowedfromIT localization, interfacetext, includingthe names of entities (e.g., ‘stratigraphic unit’), attributes (e.g., ‘soil color’), and controlled-vocabulary values (‘Munsell 5YR’), can be saved and exported using widely-shared terminology(includinguniquelyidentifiedtermsinanontology)butdisplayedusingthepreferredlanguageofanindividual project (e.g., ‘stratigraphic unit’ can display as ‘context’). Second, open-linked dataURIscanbeembeddedinallentities,attributes,andcontrolled-vocabularyvalues(linking,e.g.,speciestothe Encyclopedia of Life, or places to Pleiades). Finally, data can be systematically transformed oramplifiedduringexport,afinalopportunityformappingtosharedontologiesorlinkingtoURIs.Theseapproachesbalancetheflexibilityrequiredbyarchaeologistswiththeabilitytoproduceinteroperabledata”(Ross2015).

Similar tools are necessary for describing data recorded in laboratory work. One such tool isRightField126. The open source tool (implemented in Java) has been developed at the School ofComputerScience,UniversityofManchester(UK)togetherwithotherbioinformaticsresearchgroups(Wolstencroftetal.2011;Wolstencroft2012).RightFieldallowsscientistseasysemanticannotationofspreadsheetdatawithcommonvocabularyoftheirareaofresearchusingsimpledrop-downlists.For each annotation field, a range of allowed terms from a chosen vocabulary can be specified.Vocabulariescaneitherbe importedfroma localsystemoraregistry/repositoryofvocabularies inSKOS, RDFS or OWL (e.g. the BioPortal for biological vocabularies). The generated semanticinformation(and itsprovenance) isallheldwithinthespreadsheet.Datasharing initiativescanuseRightFieldtogenerateanddistributeaspreadsheettemplatetolaboratoryscientistsandcollectandintegratethedataandsemanticannotations.

126RightField,http://www.rightfield.org.uk




Briefsummary

Showcase examples of Linked Data applications in the field of cultural heritage (e.g. museumcollections)sofardependedheavilyonthesupportofexpertswhoarefamiliarwiththeLinkedDatamethods and required tools (often their own tools). But such know-how and support is notnecessarilyavailableforthemanyculturalheritageandarchaeologyinstitutionsandprojectsacrossEurope.AmuchwideruptakeofLinkedDatawillrequireapproachesthatallownon-ITexperts(e.g.subjectexperts,curatorsofcollections,projectdatamanagers)domostoftheworkwitheasytousetoolsandlittletrainingeffort.

Anumberofprojectshavereportedadvancesinthisdirectionbasedontheprovisionofusefuldatamapping recipes and templates, proven tools, and guidance material. For example, the STELLARLinkedData toolkithasbeenemployed in severalprojectsandappears tobeuseablealsobynon-expertswithlittletrainingandadditionaladvice.

Good tutorials and documentation of projects are helpful, but the need for expert guidance invariousmattersofLinkedOpenDataisunlikelytogoaway.Forexample,therearealotofimmature,nottriedandtestedsoftwaretoolsaround.Thereforeadviceofexpertsisnecessaryonwhichtoolsare reallyprovenandeffective for certain tasks, andprovidersof such tools shouldofferpracticaltutorialsandhands-ontraining,ifrequired.Experiencedpractitionerscanalsohelpprojectsnavigatepastdeadendsandsteerprojectteamstowardbestpractices.

Alsomore needs to be donewith regard to integrating LinkedData vocabularies in tools for datarecording in the field and laboratory. Like other researchers archaeologists typically show littleenthusiasm to adopt unfamiliar standards and terminology, which is perceived as difficult, time-consuming,andmaynotofferimmediatepracticalbenefits.

Proposed tools thereforeneed to fit intonormalpracticesandhide the semanticapparatus in thebackground, while supporting interoperability when the data is being published. NoteworthyexamplesaretheFAIMSmobiledatarecordingtoolsandtheRightFieldtoolforsemanticannotationoflaboratoryspreadsheetdata.

Recommendations

o Focusonapproachesthatallownon-ITexpertsdomostof theworkofLinkedDatageneration,publicationandinterlinkingwithlittletrainingeffortandexpertsupport.

o Provide useful data mapping recipes and templates, proven tools and guidance material toenablereducingsomeofthetrainingeffortandexpertsupportwhichisstillnecessaryinLinkedDataprojects.

o SteerprojectstowardsLinkedDatabestpracticesandprovideadviceonwhichmethodsandtoolsarereallyprovenandeffectiveforcertaindataandtasks.

o CurrentpracticesareverymuchfocusedonthegenerationofLinkedDataofcontentcollections.More could be done with regard to integrating Linked Data vocabularies in tools for datarecordinginthefieldandlaboratory.



6.4 PromoteKnowledgeOrganizationSystemsasLinkedOpenData

Knowledge Organization Systems (KOSs) such as ontologies, classification systems, thesauri andothers are among themost valuable resources of anydomainof knowledge. Becauseof the largevarietyofculturalartefactsandcontextstheculturalheritagesectorisparticularlyrichinKOSs.Intheweb of Linked Data KOSs are infrastructural components which provide the conceptual andterminologicalbasis forconsistent interlinkingofdatawithinandacross fieldsofknowledge. Theycan serve as bridges which enable interoperability between dispersed and heterogeneous dataresources. Therefore KOSs should be openly available and of course in appropriate Linked Dataformats.

MostLinkedOpenDataKOSsarebeingdevelopedfromexistingsystems.Thedevelopmentrequirescollaboration of domain and technical experts, or domain experts with the required mix ofknowledgeandskills.AsJohnUnsworthonceputitforKOSsingeneral,“Insomeform,thesemanticweb is our future, and it will require formal representations of the human record. Thoserepresentations–ontologies,schemas,knowledgerepresentations,callthemwhatyouwill–shouldbeproducedbypeopletrainedinthehumanities.Producingthemisadisciplinethatrequirestraininginthehumanities,butalsoinelementsofmathematics,logic,engineering,andcomputerscience.Uptonow,mostofthepeoplewhohavethismixofskillshavebeenself-made,butaswebecomeseriousaboutmakingtheknownworldcomputable,wewillneedtotrainsuchpeopledeliberately.Thereisagreatdealofwork forsuchpeopletodo–notallof it technical,byanymeans.Muchof thismap-makingwillbesocialwork,consensus-building,compromise.Buteven thatwillneed tobedonebypeople who know how consensus can be enabled and embodied in a computational medium.Consensus-based ontologies (in history, music, archaeology, architecture, literature, etc.) will benecessary,inacomputationalmedium,ifwehopetobeabletotravelacrossthebordersofparticularcollections,institutions,languages,nations,inordertoexchangeideas”(Unsworth2002).

6.4.1 KnowledgeOrganizationSystems(KOSs)

Knowledge organization systems (KOSs) can take different forms, e.g. glossary, thesaurus,classificationscheme,ontology(Souzaetal.2012;Bratková&Kučerová2014).AKOSmaybeusedbyinstitutionsinmanycountries,mainlyinonecountryorasa“home-grown”vocabularyonlybyoneinstitution.MostKOSsarebeingusedascontrolledvocabulariestoselectpreferredterms,namesorother “values” for certain fields ofmetadata records. For example, a subjects thesaurus providestermsforthesubjectsofdocumentsoragazetteerprovidesnamesandgeo-coordinatesforplaces.An ontology provides a conceptual model of a domain of knowledge (e.g. the CIDOC ConceptualReferenceModel).

Some years agomanyKOSswere stillmade available as copyrightedmanuals in PDF format or assimpleonline lookuppages.Recentlyopen licensingofKOSshasbecomethenormandevermoreexistingKOSsarebeingpreparedandpublishedasLinkedOpenDataforotherstore-use.

The RDF family of specifications provides “languages” for KOSs such as Simple KnowledgeOrganizationSystem(SKOS),RDFSchema(RDFS)andWebOntologyLanguage(OWL).Therelativelylightweight language SKOS127 can be used to transform a thesaurus, taxonomy or classificationsystemtoLinkedData;itcanofcoursealsobeusedtobuildanewKOS,ifnecessary.ReleasedasaW3Crecommendationin2009,thelanguagehasbeenadoptedbymanyKOSowners/developersto

127W3C(2009)Recommendation:SKOSSimpleKnowledgeOrganizationSystem,18August2009,

https://www.w3.org/2004/02/skos/



transform (“SKOSify”) controlled vocabularies for use in the web of Linked Data. KOSs that arecomplex conceptual reference models (or ontologies) of a domain of knowledge are typicallyexpressedinRDFSchema(RDFS)128ortheWebOntologyLanguage(OWL)129.

KOSs in the mentioned languages are machine-readable which allows various advantages. Forexample a SKOSified thesaurus employed in a search environment can enhance search & browsefunctionality (e.g. facetted search with query expansion), while Linked Data ontologies can allowautomatedreasoningoversemanticallylinkeddata.

6.4.2 Culturalheritagevocabulariesinuse

Before looking intothedevelopmentofculturalheritageandarchaeologicalKOSsasLinkedData itwillbegoodtohaveaviewonthecurrentusedofKOSsinthesefields.ForculturalheritageastudyoftheAthenaPlusprojectgivesanimpression,andforarchaeologythevarityofvocabularyusagebyARIADNEdatapartnersmaybeindicativeforthesituation.

AthenaPlusstudyofvocabulariesinuse

AthenaPlus(2013a)collectedandanalysedinformationon52culturalheritagevocabulariesthatareinuseat33organisationsinEurope.Themainresultsofthestudycanbesummarisedasfollows:

o Most of the vocabularies are thesauri or classification systems with a more or less complexhierarchical structure. Some are flat lists of terms which may combine terms from differentterminologies.

o Mostoftheorganisationsuseanownvocabularydevelopedin-house,oftenwithnoreferencetostandards(e.g.ISOthesauristandards)130;thisgroupincludesnational-levelorganisations.

o Multi-lingual vocabularies are rare, only a few vocabularies have concepts in more than onelanguage.

o Thevocabulariesaremainlyusedforindexingandasaqueryfeatureofanonlinedatabase.

o Mostvocabularieshaveuniqueidentifiersfortheconcepts,andonlyfewmanagementsystemsdonotallowtoexportthemfromthelocaldabase(e.g.inaCSV-file).

o The situation concerning copyrights (licensing) is varied, some vocabularies are free of rights,some organisations apply a Creative Commons license, others have not sought to clarifycopyrightsyet.

Someof thevocabulariesmaybeusedbyarchivesandmuseumsthatholdarchaeologicalartifactsamongotherculturalheritageobjects,butfewseemtoberelevantforarchaeologicalresearchdatasetsduetolackofspecifictermsforthisdomain.

VocabularyusebyARIADNEpartners

The pattern of vocabulary use by ARIADNE data partners is roughly similar to the results of theAriadnePlusstudy(cf.ARIADNE2013):128W3C(2014)Recommendation:RDFSchema1.1,25February2014,http://www.w3.org/TR/rdf-schema/129W3C(2012)Recommendation:OWL2WebOntologyLanguageDocumentOverview(SecondEdition),11

December2012,https://www.w3.org/TR/2012/REC-owl2-overview-20121211/130ISOthesauristandards:ISO2788:1974/1986(monolingual),ISO5964:1985(multilingual),orISO25964-

1/2:2011(thesauriandinteroperabilitywithothervocabularies).



o Threepartnersuseinternationaland/ormulti-lingualvocabularies(morethantwolanguages):

- EuropeanLanguageSocialScienceThesaurus(ELSST)131,

- General Multilingual Environmental Thesaurus (GEMET)132 and part of the Tree of Lifetaxonomyforwoodspecies133,

- PACTOLSthesaurus(multi-lingual)134.

o Fourpartnersusenationalstandardvocabularies

- GeologicalSurveyof Ireland (classifications forgeology,petrologyandsoils)135,PlacenamesDatabase of Ireland136, Irish National Monuments Service monument class list137, Artefactclassification138,

- SwedishMonumenttypevocabulary139,

- ArcheologischBasisregister(ABR,Netherlands)140,

- PICOthesaurus141andSITARvocabularies(Italy)142.

o Sevenpartnersuseproprietarycontrolledvocabularies(thesauri,termlists),

o Threepartnerscurrentlydonotusecontrolledvocabularies.

SomeofthevocabulariesmentionedarealreadyavailableinSKOS(e.g.GEMETsincemanyyears)orsuchaversionisinpreparation(seebelow).

6.4.3 DevelopmentofKOSsasLinkedOpenData

The firstgenerationof culturalheritageSemanticWebprojects (startedabout15yearsago)oftenusedmajorvocabulariessuchastheGettythesauri,Iconclass(NetherlandsInstituteforArtHistory)andothersfor“researchpurposes”,i.e.withoutallowancetosharepubliclyvocabularyLinkedData131ELSSTisabroad-based,multilingualthesaurusforthesocialsciences.Itiscurrentlyavailablein12

languages:Czech,English,Danish,Finnish,French,German,Greek,Lithuanian,Norwegian,Romanian,SpanishandSwedish,http://elsst.ukdataservice.ac.uk

132GEMET(EIONET/EuropeanEnvironmentAgency),http://www.eionet.europa.eu/gemet/133TreeofLife(TOL)project,http://tolweb.org/tree/134PACTOLS-Peuples,Anthroponymes,Chronologie,Toponymes,Oeuvres,LieuxetSujets(Fédérationet

ressourcessurl’Antiquité(FRANTIQ,France),http://pactols.frantiq.fr135GeologicalSurveyofIreland,http://www.gsi.ie136PlacenamesDatabaseofIreland,http://www.logainm.ie/en/137IrishNationalMonumentsServicemonumentclasslist,

http://webgis.archaeology.ie/NationalMonuments/WebServiceQuery/Lookup.aspx138NationalMuseumofIreland:Artefacts,http://www.museum.ie/en/list/artefacts.aspx139Seehttp://www.fmis.raa.se(lämningstyp)andSwedishNationalHeritageBoard(2014),extendedbythe

SwedishNationalDataService(SND)withkeywordsresearchersusewhendepositingdatawithSND.140ArcheologischBasisregister(CulturalHeritageAgencyoftheNetherlands),

http://cultureelerfgoed.nl/dossiers/archis-30/archeologisch-basisregister-plus141PICOthesaurus(CentralInstitutefortheUnionCatalogue-ICCU,Italy;termsinItalianandEnglish,butnot

archaeology-specific),http://purl.org/pico/thesaurus_4.2.0.skos.xml142SITARProjectDataModel&DataSet(SoprintendenzaSpecialeperiBeniArcheologicidiRoma),

https://www.academia.edu/5029017/MiBACT-SSBAR_SITAR_Project_Data_Model_presentation_at_the_ARIADNE_Workshop_in_Pisa_7-8.11.2013_



theyproduced frompartsof such resources.Themove toOpenandLinkedDatavocabularieswasinitiated by the library community, for example the US Library of Congress (since 2009)143, OCLC(worldwide librarycooperative)144andothers. Inrecentyearstheownersofmajorvocabularies forthehumanitiesandculturalheritagefollowed.

In 2012 Iconclass, the widely used classification system for visual content of cultural works (e.g.iconography),wasmadeavailableasLinkedOpenData145.In2014/2015theGettyResearchInstitutereleased three of their vocabularies as Linked Open Data: Art & Architecture Thesaurus (AAT),ThesaurusofGeographicNames(TGN)andUnionListofArtistNames(ULAN);theCulturalObjectsNameAuthority(CONA)was intendedtofollowinFall2015butseemstorequiremoreeffortthanexpected.146

In theUKtheSENESCHALproject (2013-2014)147 transformedseveralculturalheritagevocabulariesof English Heritage, Royal Commission on the Ancient and Historical Monuments of Scotland(RCAHMS)andRoyalCommissionontheAncientandHistoricalMonumentsofWales(RCAHMW)toSKOS and made them available online148 (Binding & Tudhope 2016). SENESCHAL built on theexperience and tools developed in the STAR and STELLAR projects (2007-2011)149. The goal of theprojectwas tomake iteasier forvocabularyproviders topublish theirvocabulariesasLinkedDataand forusers to index theirdatawithuniquely identified termsof theSKOSifiedvocabularies.TheprojectdevelopedRESTfulwebservices that facilitateconceptsearching,browsing, suggestionandvalidation. Furthermore browser-based widgets (predefined user interface controls) are availablethat allow for embedding the vocabularies inweb pages andweb forms to better index data andimprovesearchapplications.

ManyothershavealsoalreadytransformedtheirvocabulariestoSKOSordevelopednewonesbasedonthestandard.Someexamplesrelevantforarchaeologicaldataare:ThePACTOLSthesaurus150ofthe Fédération et ressources sur l’Antiquité (FRANTIQ), France, is a multi-lingual thesaurus thatfocusesonantiquityandarchaeologyfromprehistorytotheindustrialage(termsinFrench,English,German,Italian,Spanish,Dutch,andsomeArabic).

IntheNetherlandstheRijksdienstCultureelErfgoed(CulturalHeritageAgency)haveproducedSKOSversionsoftheirArcheologischBasisregister(ABRr+)andotherthesauri151.Someofthemhavebeenused in ARIADNE to explore the extraction of (meta-)data fromDutch fieldwork reports based on

143LibraryofCongress:LinkedDataService,http://id.loc.gov;LibraryofCongressSubjectHeadings(LCSH),

MARCCodeLists,ThesaurusofGraphicMaterials,AFSEthnographicThesaurusandothers.144OCLC(worldwidelibrarycooperative):LinkedData,http://oclc.org/developer/develop/linked-data.en.html;

available:DeweyDecimalClassification(DDC),VirtualInternationalAuthoritiesFile(VIAF),FacetedApplicationofSubjectTerminology(FAST)andWorldCat.

145IconclassasLinkedOpenData,http://www.iconclass.org/help/lod146GettyVocabulariesasLinkedOpenData,http://www.getty.edu/research/tools/vocabularies/lod/index.html147SENESCHAL-SemanticEnrichmentEnablingSustainabilityofArchaeologicalLinks(UK,AHRC-fundedproject,

2013-2014),http://hypermedia.research.southwales.ac.uk/kos/seneschal/148HeritageData,http://www.heritagedata.org149STAR-SemanticTechnologiesforArchaeologicalResources(UK,AHRC-fundedproject,2007-2010),

http://hypermedia.research.southwales.ac.uk/kos/star/;STELLAR-SemanticTechnologiesEnhancingLinksandLinkedDataforArchaeologicalResources(UK,AHRC-fundedproject,2010-2011),http://hypermedia.research.southwales.ac.uk/kos/stellar/

150PACTOLS(Peuples,Anthroponymes,Chronologie,Toponymes,Œuvres,LieuxetSujets),http://pactols.frantiq.fr

151RijksdienstCultureelErfgoed:Erfgoedthesaurus,http://www.erfgoedthesaurus.nl



namedentityrecognition(ARIADNE2015c).InSwedentheRiksantikvarieämbetet(NationalHeritageBoard)aimsto translate theirvocabularies (e.g. theSwedishmonuments types thesaurus) toSKOSand release them as Linked Open Data. This work is under way in their Digital ArchaeologicalWorkflowprogramme,2013-2018(Smith2015:219).

Examples of Linked Data vocabularies for research specialities are the Nomisma ontology fornumismatics152, the set of vocabularies for epigraphy developed by the EAGLE project153, and themulti-lingualvocabularyfordendrochronologicaldatabasedontheTreeRingDataStandard(TRiDaS)standard154.ThevocabuarlyhasbeendevelopedbyDataArchivingandNetworkedServices (DANS,Netherlands), with support by ARIADNE. The vocabulary is being employed for the DigitalCollaboratoryforCulturalDendrochronology155(Jansma2013)andavailablealsotootherusers.

Asthecaseofdendrochronologyremindsus,LinkedDatavocabulariesforarchaeologicaldataareofcoursenot limitedtoculturalartefacts.Suchvocabulariesarealsoneeded fordescribingbiologicalremainsofhumans,animalsandplants.TherearemanyrelevantbiologicalvocabulariesavailableinLinked Data formats shared on the BioPortal156, and may increasingly be used by archaeologicalinstitutionsandprojectstointegratedatasets.OneexampleisaprojectthatemployedconceptsoftheUberAnatomyOntology (UBERON)157 for zooarchaeological data (Kansaet al. 2014;Whitcher-Kansa2015).

An interesting casewhere a vocabulary of an established system is being transformed to SKOS isTAXREF, the French national taxonomic reference for fauna, flora and fungus (Callouet al. 2015).TAXREF is being used for the National Inventory of Natural Heritage (INPN)158, and theArchaeozoologicalandArchaeobotanical InventoriesofFrance(I2AF)database159(Callouetal.2009and 2011). TAXREF and the databases aremaintained by the FrenchNationalMuseumofNaturalHistory(MNHN),theI2AFincollaborationwithamulti-institutenetworkofbioarchaeologists160.

InadditiontopublishingTAXREFinSKOSitisintendedtosetupaWebserviceallowingtoquerythetaxonomyand retrieve results indifferent formats suchasXML/RDFand JSON.Furthermore there

152Nomismaontology,http://nomisma.org/ontology153EAGLEvocabularies(Material,Typeofinscription,Executiontechnique,Objecttype,Decoration,Dating

criteria,Stateofpreservation),http://www.eagle-network.eu/resources/vocabularies/154TreeRingDataStandard(TRiDaS),vocabularies:http://www.tridas.org/vocabularies/155DigitalCollaboratoryforCulturalDendrochronology-DCCD,http://dendro.dans.knaw.nl,seealso:

https://vkc.uu.nl/vkc/dendrochronology/156BioPortal(USNationalCenterforBiomedicalOntology),https://bioportal.bioontology.org157UBERON-UberAnatomyOntology(http://uberon.org)isacross-speciesanatomyontologythatrepresents

bodyparts,organsandtissuesinavarietyofanimalspecies,withafocusonvertebrates;itincludesrelationshipstotaxon-specificanatomicalontologies,allowingintegrationoffunctional,phenotypeandexpressiondata;seeMungalletal.(2012).

158InventaireNationalduPatrimoineNaturel/NationalInventoryofNaturalHeritage(Muséumnationald’Histoirenaturelle),http://inpn.mnhn.fr

159InventairesarchéozoologiquesetarchéobotaniquesdeFrance(I2AF),https://inpn.mnhn.fr/espece/inventaire/I100

160GDR3644BioArchéoDat,Sociétés,biodiversitéetenvironnement:donnéesetrésultatsdel’archéozoologieetdel’archéobotaniquesurleterritoiredelaFrance,http://archeozoo-archeobota.mnhn.fr/spip.php?article236&lang=fr



are plans to create mappings to other KOSs such as the NCBI Organismal Classification161, theGeoSpeciesontology162,theENVOenvironmentontology163,GeoNamesandothers.

TheI2AFdatabaseisbeingpopulatedwithdataonfloraandfaunafromarchaeologicalinvestigationscarriedout inFrenchterritories.Whendata fromarchaeological reports is imported into I2AF, it isaligned toTAXREFanda thesaurusof culturalperiods (theoldest recordsdateback to theMiddlePalaeolithic). In 2015 I2AF contained 180,000 data items concerning 2700 animal and 1100 plantspecies. The data was based on more than 3200 references, 85% “grey literature” such asexcavationsreports,specialiststudiesandothermaterial,referringto4700archaeologicalsitesand46,600contexts(pits,well,stratigraphicunitsetc.).

6.4.4 KOSsregistries

WiththegrowthoftheWorldWideWebsincethe1990severmoreKOSshavebeenpublishedonthe Web. Initially they were provided as text documents or simple HTTP pages for looking upvocabularyterms.MorerecentlyvocabularieswereimplementedasdatabasesinXML,andwithRDFthey can not only be published on theWeb but become part of theweb of LinkedData. Indeed,major vocabularies are important hubs in thisweb, for example, the AGROVOC thesaurus for theagricultureandfoodsector(whichisalignedwith16othervocabularies)164.TheW3CLibraryLinkedData Incubator Group envisage thatmajor vocabularies can play an important role in theWeb ofData as value vocabularies, provided that they are expressed with the unique identifiers (URIs)requiredfortheiruseinLinkedData(Isaacetal.2011).

The proliferation of KOSs (in various formats) has led to the creation of registries that provideinformation about vocabularies, relevant for one or all sectors, collected by the registry and/orsubmitted by vocabulary owners/developers (Golub & Tudhope 2009; Golub et al. 2014). As anexample of a domain registry, Agricultural InformationManagement Standards (AIMS)maintain acatalogueofvocabulariesfortheagricultureandfoodsector(about120vocabularies)165.Thelargestmulti-domain registry is theBARTOC -Basel Registerof Thesauri,Ontologies&Classifications166 ofthe BaselUniversity Library (Switzerland). The registrywas launched in 2013 and documents over1800 KOSs (Ledl & Voß 2016); it also briefly describes and links to 70 other, more specializedvocabulary registries. On BARTOC vocabularies can be searched and filtered based on severalcategories, including type, topic, language, location, access (e.g. freeor licensed), and format (e.g.CSV, XML, JSON, RDF, SKOS). For 139 vocabularies a SKOS version seems to be available (7.5% of1846entriesasof19/7/2016).

If we look for registries of KOSs in Linked Data formats specifically, there is the Linked OpenVocabularies (LOV) registry which currently documents 560 ontologies (Vandenbussche et al.2015)167. LOV does not register thesauri or other terminology resources, but general and domainontologiesinRDFSorOWL,whichothersmaywishtore-useasawholeoronlycertainclassesandproperties.Anexampleofacomprehensivedomainregistryofontologies is theBioPortal168,which161NCBIOrganismalClassification,https://bioportal.bioontology.org/ontologies/NCBITAXON162GeoSpeciesontology,https://bioportal.bioontology.org/ontologies/GEOSPECIES163EnvironmentOntology,https://bioportal.bioontology.org/ontologies/ENVO164AGROVOCLinkedOpenData,http://aims.fao.org/standards/agrovoc/linked-open-data165Vocabularies,MetadataSetsandTools(VEST)registry:KOS,http://aims.fao.org/vest-registry/vocabularies166BARTOC,http://www.bartoc.org167LOV-LinkedOpenVocabularies(LOV),http://lov.okfn.org168BioPortal,http://bioportal.bioontology.org



documentsover300biological/bio-medicalvocabulariesthatcanbebrowsedanddownloaded;theportalalsoshowsmappingsbetweenclassesindifferentontologies.

For cultural heritage and archaeology Linked Data vocabularies a comprehensive internationalregistrydoesnotexistasyet.Atthenational leveltheForumonInformationStandardsinHeritage(FISH)providesalistofBritishvocabulariesthatcanbeconsultedonlineand/ordownloadedasCSVor PDF; for nine vocabularies available in SKOS format FISH links to the Heritage Data serverimplementedbytheSENESCHALproject169. InFinlandtheFinnishOntologyLibraryService(ONKI)170includes KOSs of the cultural sector (Hyvönen, Viljanen et al. 2008; Suominen et al. 2014). In theNetherlandstheCATCHvocabularyandalignmentrepository171onceaimedtocovervocabulariesoftheculturalheritagedomain(vanderMeijetal.2010).

Atpresent it isdifficult to identifyvocabularies suchas thesauriorontologies for culturalheritageandarchaeologythatarealreadyavailableinLinkedDataformats(SKOS,RDFS,OWL)orareworkinprogress.AKOSregistrycouldhelpfindingpotentiallyrelevantvocabularyresourcesforre-useasawholeorforselectingrelevantconcepts/terms.AsLangetal.note,“Tacklingthislackofacommonrepositoryforstoringarchaeologicalvocabularieswithapersistentidentifierforeachconceptwillbeoneof themain issuesof theSKOS-community in the future” (Langetal. 2013). This issuehasnotbeensolvedasyet.Itmayalsobequestionedifitmakessensetoimplementaregistryorrepositoryspecifically for cultural heritage and archaeology Linked Data vocabularies. Maybe an availableregistryofallkindsofLinkedDataresourcesliketheDataHubisasufficientorevenbettersolution?

At this stage, arguably a solution should be preferred that supports community building ofdevelopers and users of Linked Data vocabularies. Registration is but one important function (forwhich theDataHubmaydo), but asor evenmore important is fostering a community that valueshigh-qualityandactivelycuratedvocabularies.Becausemanypublishedvocabulariesdonotconformto theLinkedDataprinciples,e.g. lackdereferencableHTTPURIs for retrievingdescriptionsofKOSconcepts/terms.Schmachtenbergetal.(2014b)foundthatof375proprietaryvocabularies(definedasbeingusedbyonlyonedataset) only 19%were fully and8%partially dereferencable, 73%hadtermURIsnotdereferencableatall.Only21%setlinkstooneormoreothervocabularies.

OnereasonfortheweaknessofproprietaryvocabulariesisthattherapiduptakeoftheLinkedDataapproach by many data providers has not been accompanied by training and support for propervocabularymodelling. Corcho et al. (2015) note a general preference of light-weight vocabularies(e.g. FOAF) and combinations thereof. Such vocabularies may be designed badly or, even, be“Frankensteinontologies”,i.e.conceptscobbledtogetherinconsistentlyfromdifferentvocabularies.ProvidingsupportforproperLinkedDatavocabularycreationthereforeisseenas“oneofthemainchallengesthattheontologyengineeringfieldwillhavetoaddress”(Corchoetal.2015:16).

In this challenge,aKOS registry could serveasan instrumentofquality control, improvementandconfirmation. Zimmermann (2010) suggested a quality assessment process for Linked Datavocabularies inwhich some criteria can be checked automatically (e.g. dereferencableURIs)whileothers require judgement by domain experts, e.g. clear labels and description of each term,adequacyofthecomplexityandgranularityoftheKOStointendeduses.

169ForumonInformationStandardsinHeritage(FISH):http://heritage-standards.org.uk/fish-vocabularies/;see

alsoHeritageData:Vocabulariesprovided,http://www.heritagedata.org/blog/vocabularies-provided/170ONKI-FinnishOntologyLibraryService(currently87KOSsofwhich13arerelevantforthedomainofculture

andculturalheritage),http://onki.fi;seealso:http://finto.fi/en/171CATCHVocabularyandalignmentrepositorydemonstrator,http://www.cs.vu.nl/STITCH/repository/



A useful feature of a KOS registry would also be that Linked Data vocabulary projects can beannounced so that duplication of work may be prevented and collaborative efforts fostered. Aregistrymayalsopromotejointactivitiessuchasvocabularyalignments,vocabulary-levellinkswhichincreasetheinteroperabilityofdatasetsbasedontermsthatarecommonacrossthem.


Briefsummary

Knowledge Organization Systems (KOSs) such as ontologies, classification systems, thesauri andothers areamong themost valuable resourcesof anydomainof knowledge. In thewebof LinkedDataKOSsprovidetheconceptualandterminologicalbasisforconsistentinterlinkingofdatawithinandacrossfieldsofknowledge,enablinginteroperabilitybetweendispersedandheterogeneousdataresources.

TheRDFfamilyofspecificationsprovides“languages”forLinkedDataKOSs.Therelativelylightweightlanguage Simple Knowledge Organization System (SKOS) can be used to transform a thesaurus,taxonomy or classification system to Linked Data. KOSs that are complex conceptual referencemodels (orontologies)ofadomainofknowledgeare typicallyexpressed inRDFSchema (RDFS)ortheWebOntology Language (OWL). LinkedData KOSs aremachine-readablewhich allows variousadvantages. For example a SKOSified thesaurus employed in a search environment can enhancesearch & browse functionality (e.g. facetted search with query expansion), while Linked Dataontologiescanallowautomatedreasoningoversemanticallylinkeddata.

SomeyearsagomanyKOSswerestillmadeavailableascopyrightedmanualsoronlinelookuppages.Recently open licensing of KOSs has become the norm and ever more existing KOSs are beingpreparedandpublishedasLinkedOpenDataforotherstore-use.Followingthepath-breakinglibrarycommunity, the initiative for KOSs as LOD is under way also in the field of cultural heritage andarchaeology. Some international and national KOSs are already available as LOD, Iconclass, Gettythesauri(e.g.Arts&ArchitectureThesaurus),severalUKculturalheritagevocabularies,thePACTOLSthesaurus(France,butmulti-lingual),andothers.

But more still needs to be done for motivating and enabling owners of cultural heritage andarchaeology KOSs to produce LOD versions and align them with relevant others, for examplemapping proprietary vocabulary to major KOSs of the domain. Also more LOD KOSs for researchspecialities,suchastheNomismaontologyfornumismatics,arenecessary.

The sector of cultural heritage and archaeology could also benefit from a dedicated internationalregistryforKOSsalreadyavailableasLODorinpreparation.AnauthoritativeregistrycouldserveasaninstrumentofqualityassuranceandfosteracommunityofKOSsdeveloperswhoactivelycuratevocabularies.Sucha registrycouldalsoallowannouncingLODKOSsprojects so thatduplicationofworkmaybepreventedandcollaborativeeffortspromoted(e.gvocabularyalignments).

Recommendations

o FostertheavailabilityofexistingKnowledgeOrganizationSystems(KOSs)foropenandeffectiveusage, i.e. openly licensed instead of copyright protected, machine-readable in addition tomanualsandonlinelookuppages.

o Provide practical guidance and suggest effective methods and tools for the generation,publicationandlinkingofKOSsasLinkedOpenData(LOD).

o Encourage institutional owners/curators of major domain KOSs (e.g. at the national level) tomakethemavailableasLOD.



o Promote alignment ofmajor domain KOSs andmapping of proprietary vocabulary, e.g. simpletermlistsortaxonomiesasusedbymanyorganizations,tosuchKOSs.

o Promotearegistry fordomainKOSsthatsupportsqualityassuranceandcollaborationbetweenvocabularydevelopers/curators.

6.5 FosterreliableLinkedDataforinterlinking

The principles for Linked Data include that publishers should link their data to other datasets. Inpracticethisprincipleisoftennotfollowed,particularlyalsonotinthefieldofculturalheritageandarchaeology. There are several reasons for this shortcoming, in the first place arguably a lack ofrelevant, high-quality and reliable other datasets.Without such resources awebof archaeologicalLinkedOpenDatawillnotemerge.Forbuildingthiswebacommunityofcuratorsisnecessarywhotakecareforpropergeneration,publicationandinterlinkingofLODdatasetsandvocabularies.

6.5.1 Currentlackofinterlinking

TheLinkedDataprinciplesaremeanttoenableanddrivethelinkingofinformationinanopen“webofdata”.Thecoreprinciple inthisregard isthatpublishersshould linktheirdatatootherpeople’sdata toprovideuserswithmorecontextandallowthemtodiscover related information (Berners-Lee’s principle 4). This principle is often not followed: In the 2014 LOD Cloud survey of the 1014identified datasets 445 (43.89%) did not set any out-gowing RDF links; they were either only thetargetofRDFlinksfromotherdatasetsorwereisolated.176datasets(17.36%)linkedtooneotherdataset,106(10.45%)totwoand287(28.30%)tothreeormoredatasets,79(7.79%)eventomorethan10(Schmachtenbergetal.2014a).

Also in the area of cultural heritage and archaeology few projects so far obey to Berners-Lee’sprinciple4,whichmeansthatalreadyproducedLinkedDataishighlyfragmented,awebofdatahasnotemergedyet.

Andread’Andrea (2012)argues that in thisarea interlinkingwithotheravailableresourceshasnotbeen considered sufficiently. He looked into six projects, three ofwhich had an archaeological orclassicalstudiesfocus,butfoundthattheydidnotprovidelinkstoadditionalexternalLinkedDataorattempted to integrate data of different domains. As one obstacle d’Andrea sees the lack of astandardisedapproachorat leastauthoritativerecommendationsonhowto implementthefourthLinked Data principle in the cultural heritage sector. For example, the CIDOC-CRM LODRecommendation forMuseumsmainly addresses URIs (Crofts, Doerr &Nyman 2011; ICOM 2011;CIDOC2012).

The lack of interlinking is confirmed by Leif Isaksen (2011) who for his dissertation surveyed 40projectswhich employed semantic technologies. The sample comprises of projects in the fields ofculturalheritage,archaeologyandclassicalstudies.Amongthe36data-focusedprojects(i.e.notonlyprovidinganontology), themajorityusedURIs toexpressdata (LinkedDataprinciple1),while justhalf also had dereferencable HTTP URIs (principle 2). 16 projects expressed their data as RDF(principle3),butjustfivelinkedtoexternalURIsaswell(principle4).(Isaksen2011:64)

Inacasestudy IsaksenalsoexploredapproachesforenhancingwithLinkedDatamethodsprojectswhichcreateddata interoperability inacentralisedandoftenclosedsystem(Isaksen2011,chapter7). He concludes that enhancementwill often be impractible because such projects typically havebeen small-to-medium scale in termsof numberof participants anddatasets. In suchprojects theeffort requiredofprojectpartners to convert andworkwithdata in theunfamiliar SemanticWeb



formatswould not comparewellwith the achievable “analytical return” on investment. A pay-offwouldonlymaterializeinadecentralizedlandscapeofOpenLinkedDatawherenetworkeffectscandriveadditionandinterlinkingofmoredatasets.

6.5.2 Whyistherealackofinterlinking?

There are several reasons for the neglect of the fourth Linked Data principle in the field ofarchaeology.Obviouslyonemajorreasonisthatonlyfewprojectssofarhaveproducedandexposedarchaeological Linked Data. Therefore the issue for archaeology is not a “needle in a haystack”problem.SomeLinkedDataresearchersassumethatthereisadifficultytoidentifyintheLinkedDataCloudresourceswhichareworthtolinkwith(e.g.Nikolov&d’Aquin2011;Nikolovetal.2012),butsuchaproblemdoesnotexistforarchaeologyandmostotherscientificdomains.

Developers of archaeological Linked Data projects will also not consider popular Linked Dataresources like DBpedia / Wikipedia as relevant candidates. But showcase examples of linking toother, scientific resources are missing or not well known. For example, the Open Context datapublicationplatformreportslinkingzooarchaeologicaldatawithEncyclopediaofLifeanimaltaxaandUberAnatomyOntology(UBERON)concepts(Kansaetal.2014;Whitcher-Kansa2015).

AndreasBlumauer(2013)thinksthatthelowlevelofexternallinkinginmostdomainsisduetotworeasons:1)thereisnotmuchdomain-specificknowledgeanddata intheLODCloud,exceptforthebiological domain (created by the Bio2RDF initiative, among others) and somehigh-quality “microLODclouds”whichhavebeendevelopedbydedicateddomainprojects;2)manydatasetsoftheLODcloud are not maintained in a professional manner and hence not trustworthy for sustainableinterlinking.FurthermoreBlumauernotesthatthereisoftenalackofclearopendatalicensing.

Smith-Yoshimura (2014c and 2016) notes a number of barriers or challenges institutionalimplementers of Linked Data services mentioned in the OCLC Research surveys 2014 and 2015.AmongthemostcitedissueswhentryingtoconsumeorlinktootherLinkedDatasetswere:

o WhatispublishedasLinkedDataisnotalwaysreusableorlacksURIs,

o Understandinghowothersdataisstructured,

o Easyaligningnotpossible(e.g.importantauthoritytermsaremissing),

o Vocabularymappingprovestobedifficult(e.grequiresalotofmanualwork,issueswithlevelofspecificityofterms),

o Lackofuseful“offtheshelf”tools(e.g.withregardtovisualisation),

o Datasetsnotbeingupdated,

o SizeofRDFdumpsandvolatilityofdataformatofdumps,

o Servicereliability,e.g.unstableSPARQLendpoints.

Otherbarriersincluded:lackofLinkedDatasetsoflocalinterest,licensesmorerestrictivethanCC-Byor ODC-BY, insufficient internal resources to incorporate available Linked Data into routineworkflows.

6.5.3 NeedofreliableLinkedDataresources

ThewebofLinkedDatawillemergefromthepublicationandinterlinkingofevermoreresourcesofdifferent providers. This means a shift from a model of single, authoritative and mostly static



metadata records to a distributed approach in which statements about items of interest (e.g.researchobjects)cancomefromdifferentresources.ThereforethequalityandcontinuedavailabilityoftheresourcesisparamountfortheoverallworkingofthewebofLinkedData.

The benefits of LinkedDatawill notmaterialize if computer applications cannot reliably use it forspecificpurposes.ButmanystudieshaveshownthatbasicLinkedDataprinciplesandadditionalbestpracticessuggestedbyleadingdevelopersareoftennotfollowed(e.g.Duanetal.2011;Hoganetal.2010;Hoganetal.2012;Schmachtenbergetal.2014a/b).

Interlinking with Linked Data of other providers requires that one can trust that their data andservicesarereliablewithregardtocriteriaofquality.HowevertheLinkedOpenDataCloudisamixof resources, someofwhichmaynot fulfil requirementswith regard to content (e.g. incomplete),othersarenotreliablewithregardtomaintenance.Buil-Arandaetal.(2013)foundthatof427publicSPARQL endpoints registered in the DataHub half were off-line and only one third were almostalwaysavailableduringamonitoringof27months.

Recent figures available from LODStats172 show that most Linked Data resources simply are notreliable. LODStats processes RDF datasets from the DataHub, data.gov and publicdata.eu datacatalogstoproducestatisticaloverviewsofthestatethedataweb(Aueretal.2012b;Ermilovetal.2016). InMay2016LODStats identified9960datasetsofwhich7112 (71.5%)presentedproblems;6712of intotal9416RDFdumpshavingerrors(71.28%)and400of intotal544SPARQLendpointswitherrors(73.53%).

Theissueofreliabilityofresourcesforlinkingisemphasisedbymanydataproviders,includingfromthe cultural heritage sector where authoritative information and well maintained services areessential.Forexampleauthorsof the librarydomainstress:“Themainproblemfor the linkeddatawebisdealingwithreliability:Isthedatacorrectanddoprocessesexistthatguaranteeahighdataquality?Whoisresponsiblefor it?Ofthesameimportance isreliability intime: Isaresourcestableenoughtobecitable,orwill itbegoneatsomepoint?Thesequestionsareofspecial importanceinthecontextof research,wherecitability isessential,and forhigher-level services thatarebasedonthiskindofdata”(Hannemann&Kett2010).

With the increasing number of Linked Data resources their quality has become a core topic ofsemantic web conference sessions and dedicated workshops. Ever more detailed schemes andmetrics for LinkedDataquality are being elaborated andused to scrutinize resources and suggestimprovements, if required (e.g.Assaf&Senart2012;Aueretal.2013 [chapter7];Behkamal2014;Fürber&Hepp2010a/band2011a173;PlanetData2012;Zaverietal.2013).Asanovelty,Hoxhaetal.(2011) base their framework on principles of “green engineering”, e.g. that it is better to preventwastethantotreatorcleanupafteritisformed.Theapproachworksparticularlywellwithregardtore-useofresourcesandalignmentwithactualuserdemand.

TheLinkedDataqualityschemestendtocentreonadherencetogoodpracticeswithregardtodataandtechnicalstandards.Butalsogeneralcriteriaarebeingaddressed,forexample,thatLDresourcesshould be easy to find and assess with regard to relevance and trustworthiness, e.g. well-documentedinageneralordomainregistry,includingdatadescription,transparentdatapolicy,dataprovenanceinformation,andothers.

172LODStats(AgileKnowledgeEngineeringandSemanticWebGroupatUniversityofLeipzig,Germany),

http://stats.lod2.eu173Seealsotherelatedwebsitehttp://semwebquality.organdtheDataQualityManagementVocabulary

(Fürber&Hepp2011b)andDataQualityConstraintsLibrary(Fürberetal.2011)



While different approaches are being used, the quality criteria essentially are about how users(humans andmachines) can discover, understand and access LinkedData resources that arewell-structured,accurate,up-to-dateandreliableovertime.IdeallytheresultofthecurrenteffortswillbeeasytousetoolsthatallowLinkedDatacuratorsmonitorresources,detectandfixproblemssothathigh-qualitywebsofdataarebeingdevelopedandmaintained.

6.5.4 FosteracommunityofarchaeologicalLODcurators

The lack of trustworthy resources in many quarters of the “web of data” makes clear a corerequirement for high-quality Linked Open Data: a community of curators who ensure reliableavailabilityandinterlinkingofLODdatasetsandvocabularies.

OnedomainofgoodLinkedDatacurationpracticeswhichcouldbe followedare theLifeSciences.Ten years ago the Life Sciences Semantic Web was described as full of “semantic creep – timid,piecemeal and ad hoc adoption of parts of standards by groups that should be stridently taking aleadershiproleforthecommunity” (Good&Wilkinson2006).Meanwhilethedomainhasadvancedsubstantially towards amore integrated area of theweb of LOD.One outstanding example is theBio2RDF174communitywhichcreatedand/orinterlinked35datasets.TheBio2RDFdatasetsareoneofthedensestclusterspresentontheLODdiagram175.

The importance of LOD curation becomes clear when considering that also a lot of life and bio-sciencesrelatedLinkedDataproducedasyetremains isolatedanddifficultto integrate.Hasnainetal.(2015)catalogued137publicSPARQLendpointsofrelevantLinkedDataprovidersandtriedtolinkconcepts and properties of the resources. They found that most resources could not be easilymappedbecausetherewasverylittlevocabularyandURIre-use,i.e.vocabularieswhichmightbridgebetweentheresourceswerenotpresent.AlsoshortcomingsofURIsarenotedasalotcouldnotbedeferencedandmanydatasetsincludedorphanURIs(i.e.“type”-lessURIinstances).

If thedomainof archaeological research aspires to growa rich and robustwebof LODwithin theoverall LOD Cloud, it will have to foster and support a community of curators who take care forproper generation, publication and interlinking of LOD datasets and vocabularies. This communitycould benefit from good practices demonstrated by the AncientWorld LOD communitymobilisedandintegratedbyPelagiosandresearchobjectcentredinitiativessuchasNomisma(seeSection5.3).


Briefsummary

The core LinkedDataprinciple arguably is thatpublishers should link their data tootherdatasets,because without such linking there is no “web of data”. In practice this principle is often notfollowed, particularly also not in the field of cultural heritage and archaeology. This means thatalreadyproducedLinkedDataremainsisolated,awebofdatahasnotemergedyet.Thereareseveralreasonsforthisshortcoming.Obviouslyonefactoristhatonlyfewprojectssofarhaveproducedandexposedarchaeological LinkedData.Developersof suchdatawill alsonot considerpopular LinkedData resources like DBpedia/Wikipedia as relevant candidates. Moreover there is the issue ofreliability,thatdataonelinkstowillremainaccessible,whichoftentheyarenot.Surveysfoundthatmanydatasetspresentproblems,forexampleSPARQLendpointsareoftenoff-lineorpresenterrors.

174Bio2RDF:LinkedDatafortheLifeSciences,http://bio2rdf.org175Cf.theLinkingOpenDataclouddiagram,http://lod-cloud.net



With the increasingnumberof LinkedData resources theirqualityhasbecomeacore topicof thedeveloper community. Detailed quality schemes and metrics are being elaborated and used toscrutinizeresourcesandsuggestimprovements.Thequalitycriteriaessentiallyareabouthowusers(humans andmachines) can discover, understand and access LinkedData resources that arewell-structured,accurate,up-to-dateandreliableover time.Furthermoretheresourcesshouldbewell-documented, e.g. with regard to data provenance and policy/licensing. Ideally the result of thequalityinitiativewillbeeasytousetoolsthatallowLinkedDatacuratorsmonitorresources,detectandfixproblemssothathigh-qualitywebsofdataarebeingdevelopedandmaintained.

The lack of trustworthy resources in many quarters of the “web of data” makes clear that acommunity of curators is necessarywho take care for reliable availability and interlinking of high-quality archaeological LOD datasets and vocabularies. A few domains already have such acommunity, the Libraries and Life Sciences domains, for instance. Also the Ancient World LODcommunityaroundthePelagiosinitiativeortheNomismacommunitycanbementionedasexamplesofgoodpractice.ItappearsthatthedomainofarchaeologyneedsaLODtaskforceandanumberofprojectswhichdemonstrateandmakeclearwhatisrequiredforreliableinterlinkingofLOD.

Recommendations

o Foster a community of LOD curators who take care for proper generation, publication andinterlinkingofarchaeologicaldatasetsandvocabularies.

o FormataskforcewiththegoaltoensurereliableavailabilityandinterlinkingofLODresources;LODqualityassuranceandmonitoringshouldbeestablished.

o Sponsor a number of projects which demonstrate the interlinking and exploitation of someexemplaryarchaeologicaldatasetsasLinkedOpenData.

6.6 PromoteLinkedOpenDataforresearch

ArchaeologicaldataandknowledgepresentagreatchallengeforLinkedData.Thischallengestemsfromthemulti-disciplinarityoftheresearchonarchaeologicalsitesandobjects(Vavliakisetal.2012).AwebofLinkedDatabasedoncross-domainanddomain-specificontologiesandterminologiescanallowaddressingbetterarchaeological researchquestions,which require integrationof knowledgeanddataofdifferentdomains.

Todaybenefitsof LinkedOpenDataaremainly framed,andsometimesdemonstrated, in termsofadvancedsearchservicesbasedonthesemantic linkingbetweenrelateddatasets.Thismayappealto cultural heritage institutions as it allowsmaking their collections better discoverable andmorerelevantbyaddingexternalcontextualinformation.

Whilesuchsearchservicesarealsoimportanttoresearchers,afocusondatasearcharguablydoesnotstronglypromotethegenerationofLinkedOpenDataofresearchdatasets.Researchgroupsandinstitutions will be much more attracted by demonstrated research dividends of semanticallyinterlinkedandintegrateddata.Suchdividendscouldforexampleresultfromcombiningdatafromseveralprojectsinwaysthatenableinterestingnewlinesofresearch,orviewsondatafromdifferentdisciplinary perspectives suggesting interdisciplinary approaches. Researchers also need effectivetools,usablebynon-ITexperts,tobenefitfromLinkedDataintheresearchprocess,e.g.exploreandexloitsemanticrelationsbetweendatasetsorbetweenpublicationsandrelateddata.

Established ways of data integration for research follow other paradigms than Linked Data. Forexampledatasharedbyresearchersinadatabasewithresearchtoolsimplementedontop,e.g.the



PaleobiologyDatabase forwhichFossilworksprovidesdataqueryandanalysis tools176.Ora stand-alone database with sophisticated modelling and interactive web interfaces such as ORBIS - TheStanfordGeospatialNetworkModeloftheRomanWorld177.ORBISallowscalculatingtheeffort(time,financial expense) associated with different types of travel in antiquity (Meeks & Grossner 2012;Scheidel2015).ApplicationsofLinkedOpenDataforresearchwillhavetodemonstrateadvantagesoverorotherbenefitsthanalreadyestablishedformsofdataintegrationandexploitation.

6.6.1 ALinkedOpenDatavision(2010)

In2010,ChristianBizer,aleadingresearcherinLinkedDatamethodsandapplications,outlineda10yearvisionfor“extendingtheWebwithaglobalscientificdataspace”(Bizer2010).BizerobservedanincreasingadoptionoftheLinkedDataapproachforsharinglibrary,governmentandscientificdata,andafirstgenerationofapplicationsthatexploitinterlinkeddatasetsfornovelinformationservices.Hisvisionforthenext10years,quotedinfull,was:

o “Linkeddatawilldevelopintothestandardtechnologyofsharingscientificdataonglobalscaleandforinterconnectingdatabetweendifferentscientificdatasources.

o TheemergingWeboflinkeddatawillcontainscientificdataaswellasdatafromotherdomainsandmightbecomeasomnipresentinourdailylivesastheclassicdocumentWebistoday.

o Most open-license scientific data setswill be directly available as linked data on theWeb. ForextremelylargedatasetsfromastronomyorphysicsforwhichitisinefficienttogenerateanRDFrepresentation, the Web of linked data will contain detailed metadata that will enable thediscoveryofthesedatasets.

o Allscientificworkenvironmentswillhavelinkeddataimportandexportfeaturesandwillprovidefor publishing scientific data directly to the Web of linked data. Disciplinary repositories ofscientificdataaswellasdataarchiveswillprovide linked-dataviewson thearchiveddataandwillthusmaketheircontentavailableontheWeb.

o ScientistswillnavigatealongRDFlinksbetweendifferentscientificdatasetsaswellasbetweenpublications and supporting experimental data. They will use linked-data search engines todiscoveralldataonglobalscalethatisrelevanttotheirquestionathand”.

Asone critical requirement for such LinkedData empowered researchBizer highlighteddiscipline-specific vocabularies (e.g. thesauri, ontologies), which need to be integrated so that a searchableweb of scientific data can emerge. Furthermore he noted that integration of LinkedData tools inscientificworkenvironmentswasmissing. So farBizer’s vision isnot realised,buthas four furtheryearstomaterializeuntil2020.

6.6.2 LODforresearch:Thecurrentstateofplay

Efforts for cultural heritage LOD so far have been investedmainly on publishing variousmuseumcollections, often linked to DBpedia/Wikipedia. Concerning special collections an outstandingexample is the numismatics databases that participate in the Nomisma initiative178. Also a few176Fossilworks,http://fossilworks.org177ORBIS-TheStanfordGeospatialNetworkModeloftheRomanWorld,http://orbis.stanford.edu178Nomisma,http://nomisma.org/datasets;severalcoindatasetsoftheAmericanNumismaticSocietyand

institutionsinEuropehavebeenmadeavailableinRDFformat;theNomismaprojectalsoprovidesanontologyfordescribingcoins.



archaeological datasets have been published as Linked Data, for example, in the STELLAR projectLinked Data of project archives deposited with the Archaeology Data Service (ADS)179. Specialmention deserves that the Getty Research Institute has published their major cultural heritagethesauri as LOD180, and also other widely employed international and national vocabularies havebecomeavailableasLOD,e.g.Iconclass181,UKthesaurimadeavailablebytheSENESCHALproject182,thePACTOLSthesaurus183,andothers.

Thelast10yearshaveseensubstantialadvancesinLODknow-how,i.e.whatisrequiredtoproduce,publishandinterlinkLODofarchaeologicalandculturalheritagecollections/databases(cf.Hyvönenet al. 2005; Aroyo et al. [eds.] 2007; Kollias & Cousins [eds.] 2008; Isaksen 2011; Tudhope et al.2011b;Elliottetal.2014;Mayetal.2015). Intotal,however,notmanydomainLODdatasetshavebeenproducedandeffectivelyinterlinkedasyet.

Ifthereisasubstantialfurther increaseinpublishedandinterlinkedLODdatasets,semanticsearchand browse applications will allow discovery and retrieval of related content/data. But such anadvance will mainly concern data aggregation, search and access, use of LOD for other researchpurposes is not implied. By use for research purposes we mean capability to address researchquestions and validate or scrutinize knowledge claims. The lack of such capability has not goneunnoticedbyresearchersanddatamanagerswhoexpectrelevanceoftheLODapproachalsointhisdirection.

For example a researcher who tried using museum Linked Data sets for an art historical studysuggests cultural heritage institutions “to seek out research uses of their data, and not limit theirthinking to mere aggregation and dissemination (…). Creating LOD is hard enough for theseinstitutions,sowithsomemoreutilitiesfor individualresearcherstotakeadvantageofthecomplexdataexpressionsandqueriesofferedbyLOD,hopefullyitwillbeeasierforGLAMstodesigntheirdataofferingstobettersupport thekindofdetailedresearchthatthesedataprojectskeeppromisingtoenable”(Lincoln2016[note:GLAMSisanacronymforGalleries,Libraries,ArchivesandMuseums]).

ARIADNE colleagueswith regard to employing the LOD approach in archaeology note: “Importantthattheseconceptsandtechnologiescontinuetobedeveloped,butthenextfiveyearsreallyneedtostartshowingitsusefulnessforansweringresearchquestions.Forexample,usingtheLDcreatedbythePortableAntiquityScheme,theBritishMuseumandADS,andlookatwhatwecanactuallylearnby combining these datasets. Are they even compatible? What makes datasets compatible forinteroperability? How compatiblemust they be in order to generate new and useful information?Doesinteroperabilityactuallyconfoundtheresults,aswedon’tunderstandhowbesttofilterit?It’sone thing to keep putting LOD out there, butwe need to partner in a focussedwaywith domainexpertstostartansweringthesequestions,beginbuildingbestpracticeonhowtoactuallyuseLD”(J.Charno,H.Wright and J. Richards,ADS, statement in the consultationon theARIADNE innovationagenda).

179ArchaeologyDataService:TheSTELLARproject,http://archaeologydataservice.ac.uk/research/stellar/;ADS

LinkedOpenData,http://data.archaeologydataservice.ac.uk180GettyVocabulariesasLinkedOpenData,http://www.getty.edu/research/tools/vocabularies/lod/;ARIADNE

usestheirArt&ArchitectureThesaurusforintegratingsubjectsrelatedinformation.181ICONCLASSasLinkedOpenData,http://www.iconclass.org/help/lod182HeritageData-LinkedDataVocabulariesforCulturalHeritage,http://www.heritagedata.org183PACTOLS-Peuples,Anthroponymes,Chronologie,Toponymes,Oeuvres,LieuxetSujets,

http://frantiq.mom.fr/thesaurus-pactols



AlsoresearchersofthedatapublicationplatformOpenContextemphasise,“ArchaeologistsneedtoseemoredirectresearchapplicationsinordertobetterjustifytheaddedcostandeffortrequiredtopublishLinkedOpenData”(Kansa&Whitcher-Kansa2013:9;seealsoKansa2015).OpenContexthasbeenworkingonprojectswith researchers and institutions that involve LinkedData. For example,one project focused on zooarchaeological datasets documenting early agricultural communities inAnatolia. The datasets have beenmade comparable by linking and annotating them according toanimal taxa published by the Encyclopedia of Life184 and to morphological concepts of the UberAnatomy Ontology185 (Kansa et al. 2014; Whitcher-Kansa 2015). This is a rare example wherearchaeologicaldatahasbeeninterlinkedwithascientificKOS,althoughnotsupportingresearchtasksbeyondsearchingobjects.

Theneedtoprogress fromLODbasedcontent/datasearchtoresearch-focusedapplications isalsostressedbythee-scienceandlinkedsciencecommunitiesthatwanttoseeLODsupporttheprocessofresearch,includingscientificworkflows,computingandanalysis(Bechhoferetal.2011;Kauppinenet al. 2013). Indeed, novel LOD based models and applications that demonstrate considerableadvances in research processes and outcomes may be decisive in fostering uptake of the LODapproachbyresearchcommunities.

6.6.3 Searchvs.research

Some examples will be useful to illustrate the difference between searching archaeologicalinformationbasedonLODandresearch-focusedLODapplications.TheGettyResearchInstitutehasmadeavailable theirmajorculturalheritage thesauriasLOD.PatriciaHarpring,ManagingEditoroftheGettyVocabularyProgram,describesascenariowherethesevocabularieswouldaiddiscoveryofrelatedinformation:

“Let’simaginethataresearcherfindsaninterestingarticleonlineaboutthehistoricaluseofincenseburnersinMexico.Toexplorethetopicfurthertodaywouldrequiremanyhoursordaysofresearch;however, LOD will enable a new generation of search engines to follow the links between datasources to deliver more complete answers in much less time. In this use case, the AAT [Art &Architecture Thesaurus] could provide variant spellings, synonyms in other languages for ‘incenseburners,’ and the narrower concept ‘censers’ with its variant terms, enabling the researcher toinstantaneouslydiscovernumerousmuseumsitesandarticlesonthistopic.TheAAThierarchycouldalso focus the search on censers attributed to Pre-Columbian cultures. The user could exploregeographicregionswherethesecenserswerecreatedthroughTGN[ThesaurusofGeographicNames]placenames,hierarchies,andlinkedmaps.ThenamesandbiographiesinULAN[UnionListofArtistNames] could lead theuser topertinent informationaboutartistsandpatronsassociatedwith thecreation of the censers. CONA [Cultural Objects Name Authority], which ideally will have subjectindexing,couldprovidelinkstophotographs,paintings,orevenYouTubevideosportrayingusageofcensers(seeanentertainingvideoofa‘monstercenser’atSantiagodeCompostela,Spain)”(Harpring2014).

Achieving this scenario for a lot of cultural heritage informationwould be a great advance in thediscoveryofrelatedinformation.AsHarpringnotes,itwouldallowfindingmorecompleteanswerstosearchquestionsinmuchlesstime.However,thisisaboutsearch,notresearch.

Beck (2010) addresses future research-focusedarchaeological applicationsof LOD.Oneexample issequencesofpotterystyleswhicharebeingusedtoestablishaframeworkfordatingarchaeological184EncyclopediaofLife,http://eol.org185UBERON-UberAnatomyOntology,http://uberon.org



contexts, e.g. stratigraphic layers of an excavation. Beck envisions that interlinked LOD of potteryclassifications and documentation of excavations would allow identifying inconsistencies in thepublishedarchaeologicalrecord.

“In addition to many other things pottery provides essential dating evidence for archaeologicalcontexts. However, pottery sequences are developed on a local basis by individualswith imperfectknowledge of the global situation. This means there is overlap, duplication and conflict betweendifferent pottery sequences which are periodically reconciled (…). This is the perennial process oflumping and splitting inherent in any classification system. Updated classifications and probabledates allowus to re-examineour existing classifications.One can reasonover thedata to findoutwhichcontexts,relationshipsandgroupsareimpactedbyachangeinthedatingsequenceseitherbyproxyorbylogicalinference(achangeinthedateofacontextproducesalogicalinconsistencywithastratigraphically related group). (…) Publicly depositedRDFdata shouldbe linkeddata: thismeansthatall theprimarydataarchivesare linked to their supporting knowledge frameworks (suchasapotterysequence).Whenaknowledgeframeworkchangestheimplicationsarepropagatedthroughtotherelateddatadynamically”.

This scenario is very demanding as it includes machine-based reasoning over LOD potteryclassifications interlinkedwith information inmanydatasetsofexcavationswhichcontaindatingofstratigraphiclayersofexcavationsbasedonpotteryfinds.Thepotteryclassificationsystem(or,morelikely,differentsystems)wouldhavetobeavailableasLinkedData(basedonSKOSorOWL),andthepotterybaseddatingsintheexcavationdatasetsdescribedconsistentlyinacommonformat,andthedatasetsofcoursealsopublishedasLinkedData.

While unrealistic, the scenario touches upon crucial issues of stablility and change of knowledgeframeworks. If they are “living” frameworks that support the on-going research and knowledgecreationprocess,thereisalwayssomeadditionandmodificationgoingon.Oneextremeexampleisspecies taxonomies where revisions are conducted regularly and produce more or less intensive“revisionshocks”whichimpactonthedocumentationofspeciesandevencriticalmeasuressuchasspecies protection and conservation (Vences et al. 2013). Hepp (2007) addresses conceptualdynamics in domains of knowledge and the issue of long update cycles of formalized knowledgeorganizationsystems.Thusnewandarguablymostinterestingconceptsincurrentresearchwillnotbe present for long in domain thesauri or ontologies. Furthermore there is the issue of differentclassificationsofthesameresearchobjectswhich, ideally,wouldco-exist inaknowledgesystemorinterlinkedsystems(cf.Madsen2004:41,inthecontextofarchaeologicalreferencecollections).

Visions of research-focused archaeological applications of LOD, like Beck’s example, expect suchapplications to allow automatic reasoning over a web ofmany interlinked data resources. In thisquasi artificial intelligence scenario Linked Data applications would identify inconsistencies,contradictions, etc. in scientific statements (knowledge claims) or, as a positive example, presentsurprisingrelationshipsbetweendataworthexploringfurther.ThusLinkedDataapplicationswouldcarryoutsometasksthatcanbesubsumedunderresearchratherthansearch,e.g.detectrelevantrelationshipsbetweendataorscientificstatementsthatarecontradictory.

6.6.4 Examplesofresearch-orientedLinkedDataprojects

TherearealreadysomeLinkedDataprojectswhichaimtogobeyondsimplesearchfunctionality.Butnotmanyandnotnecessarily inarchaeology.Wedescribe twoexamples,one in the fieldof socialhistoryandanotherconcerningClassicalStudies.



DutchShipsandSailors186:AsanexampleofLOD in the fieldof socialhistory, theDutchShipsandSailorsprojecthasbroughttogetherfourdatasetsonDutchmaritimehistoryasfive-starLinkedData.End of March 2014 the Linked Data comprised of 25 million RDF triples, divided over 33 namedgraphs. Around 1.5million links connected the datasets as well as linked to external sources; forexample 180,000 links to external historical newspaper articles were established and 2500geographical entitiesmatched to GeoNames entities (De Boer et al. 2014 and 2015). The projectpresentedanumberofexamplesofhowthedatacanbeusedforhistoricalresearchonthesocio-economicrealitiesofthe18thCentury,forexamplelistsofpersonswhoembarkedondifferenttypesofships,analysisof thebirthprovincesofsailorsonDutchEast IndiaCompanyshipsovermultipleyears, etc. In a follow-up project further datasets have been added to the initial Dutch Ships andSailorsCloud(deBoer&Leinenga2014;Entjes2015).

EPNet Project187: Aims to provide historians with data resources and tools for investigating theRomantradesystembasedonLatinandGreekinscriptionsonamphorasforfoodtransportation.Incollaboration with experts of the history of the Roman economy the project has specified anontologyofdomainknowledgewhichrepresentsthewaythedataarebeingunderstoodbyscholars,howtheyareconnected,andhowtheyrelate to the literatureandcurrent researchpractices.ThemainsectionoftheontologyisaspecialisationoftheCIDOCCRMwhileothersectionsbuildonthemetadatamodeloftheEAGLEproject(EAGLE2015),EpiDoc188fortheencodingofeditionsofancienttexts/documents (inscriptions, papyri, manuscripts), FaBiO189 for bibliographic references, andothers.TheEPNetontologyismeanttobe“functionaltoresearch”,e.g.supportresearchers intheexplorationofhypothesesandquestionestablishednarratives (Calvaneseetal.2015;Calvaneseetal. 2016). Initial data resources are the rich database of Roman amphorae and their associatedepigraphy (i.e. stamps and tituli) of the Centre for the Study of Provincial Interdependence inClassical Antiquity, University of Barcelona190, the Epigraphic Database Heidelberg191, and thePleiadesgazetteerandgraphofancientplaces192.

6.6.5 CIDOCCRMasabasisforresearchapplications

Expectations of reseach-focused applications of LOD in the field of archaeology andother culturalheritage research often relate to the CIDOC CRM as an integrating framework. Oldman (2012)explainsthattheLinkedDatapublicationoftheBritishMuseumonlinecollectiondatainCIDOCCRMformat “comes from a concern that many Semantic Web / Linked Data implementations will notprovide adequate support for a next generation of collaborative data centric humanities projects.They may not support the types of tools necessary for examining, modelling and discoveringrelationships between knowledge owned by different organisations at a level currently limited tomorecontrolledandlocalizeddata-sets”.TheResearchSpaceproject193(ledbytheBritishMuseum)isdeveloping an online collaborative environment for humanities and cultural heritage informationsharingandresearchthatbuildsonCIDOCCRMbasedmethods.186DutchShipsandSailors(ClarinIVproject,4/2013-3/2014),http://dutchshipsandsailors.nl187EPNet-ProductionandDistributionofFoodduringtheRomanEmpire:EconomicandPoliticalDynamics

(ERCAdvancedGrantproject,3/2014-2/2019),http://www.roman-ep.net188EpiDoc:EpigraphicDocumentsinTEIXML,http://epidoc.sf.net189FaBiO-FRBR-alignedBibliographicOntology,http://vocab.ox.ac.uk/fabio190CEIPACdatabase,http://ceipac.ub.edu191EpigraphicDatabaseHeidelberg,http://edh-www.adw.uni-heidelberg.de192Pleiades,http://pleiades.stoa.org193ResearchSpace,http://www.researchspace.org



Oldman(2012)alsonotesthatsincesomeyearstheCIDOCCRMhasbeenadoptedbymanyprojects“butithasalsoreacheda‘chickenandegg’stageneedingtheimplementationofpublicapplicationstoclearlydemonstrate itsuniquepropertiesandvalue tohumanities research”. This isaboutmorethansemanticsearchofrelatedcontent/databasedontheCIDOCCRMorotherontologies.

TheCIDOCCRMisintendedtoenableexchangeandintegrationofscientificdocumentationoffinds,sites andmonuments, at the level of detail and precision required by researchers of the heritagesciences194. Recent extensions of the CIDOC CRM cover scientific observation and argumentation(CRMsciandCRMinf).ThusCIDOCCRMbasedmodellingofscientificprocessesanddocumentationof observations can enable integration of scientific information and argumentation (knowledgeclaims).

The CIDOC CRM developer community invites data sharing and integration projects to use theontology to describe the meaning and context of their information objects so that research e-infrastructureandservicescanprovidehomogeneousaccesstotheinformation,inawaythatretainsits original meaning and proper context. The proponents argue that this is the way forward torelevantheritageresearchapplications.What theyseeas inadequate is thetraditional informationaggregation and integration approach based on fixed “core” metadata fields which are artificialgeneralizationsthatdonotmediatethecontextualknowledgeofthedataproviderssuchasresearchinstitutesandmuseums(Doerr&Oldman2013;Oldmanetal.2014).

The vision of the CIDOC CRM developer community goes well beyond enabling cultural heritageinstitutionstoprovidestructuredaccesstocollectionobjects.Archaeologicalandotherheritagedatacollections/databasescontainamultitudeoffactsthathavebeenestablishedwithvariousmethodsand in different contexts of research. Therefore a common way to describe the information isrequiredthatallowssemanticintegrationandaddressingquestionsbeyondthelocalcontextofdatacreationanduse.

Thisobjectivehasbeenaddressedby thedevelopmentof theARIADNEReferenceModelwhich isbased on the CIDOC CRM and enhanced or new extensions (e.g. CRMarchaeo for archaeologicalexcavations)195. The aim of semantic integration of research data requires that the participantsproduce a conceptual mapping of their database structures to the extended CIDOC CRM. ThemappingenablestheconversionandexportofthedatabasesinaCIDOCCRMcompatibleRDFformatwhichcanbesharedasLinkedDataontheWeb.

The challenge of enabling effective mappings has been addressed by an innovative solution, theSYNERGY ReferenceModel (Doerr et al. 2014b). SYNERGY is intended as a modular environmentcomposed of different instruments which will perform individual tasks of the mapping process,includingalsoa knowledgebaseof re-useablemapping cases. SeveralARIADNEhavealreadyusedtheMappingMemoryManager196moduleofSYNERGYtodefinecomplexcorrespondencesbetweenentities of their and other databases and the conceptual classes provided by the extended CIDOCCRM(ARIADNE2016a;Doerretal.2016;Gerthetal.2016).

At large scale this approachwill allow reaping the expected benefits only in themedium to longterm,whenmanydatabasesaremappedtotheextendedCIDOCCRM.However,mappingofafewrelateddatabasesmaydemonstrate significantadvantagesofCIDOCCRMbased integration in theshort-term,possiblypromotingfurthermappings.

194Cf.DefinitionoftheCIDOCConceptualReferenceModel.Version6.1,February2015,pagesi-ii,

http://www.cidoc-crm.org/docs/cidoc_crm_version_6.1.pdf195SeetheoverviewanddescriptionoftheCIDOC-CRMextensionsat:http://www.ics.forth.gr/isl/CRMext/196MappingMemoryManager-3M(FORTH-ICS),http://www.ics.forth.gr/isl/3M




Briefsummary

LinkedOpenDatabasedapplicationsthatdemonstrateconsiderableadvancesinresearchprocessesand outcomes could be a strong driver for a wider uptake of the LOD approach in the researchcommunity.CurrentexamplesofLinkedDatauseforresearchpurposesrarelygobeyondsemanticsearch and retrieval of information. This has not gone unnoticed by researchers who expectrelevanceofLinkedOpenDataalsoforgeneratingandvalidatingorscrutinizingknowledgeclaims.Toallowforsuchusesatighterintegrationofdiscipline-specificvocabulariesandeffectiveLinkedDatatoolsandservicesforresearchersarerequired.

Expectationsofreseach-focusedapplicationsofLODinthefieldofculturalheritageandarchaeologyoften relate to the CIDOC CRM as an integrating framework. The CIDOC CRM is recognised as acommon and extendable ontology that allows semantic integration of distributed datasets andaddressingresearchquestionsbeyondtheoriginal,localcontextofdatageneration.Notably,intheARIADNE project several extensions of the CIDOC CRM have been created or enhanced, e.g.CRMarchaeo,anextensionforarchaeologicalexcavations,andextensionsforscientificobservationsandargumentation(CRMsciandCRMinf).

Tomeet expectations such as automatic reasoning over a largeweb of archaeological datamanymore(consistent)conceptualmappingsofdatabasestotheCIDOCCRMwouldbenecessary.LinkedData applications then might demonstrate research dividends such as detecting inconsistencies,contradictions, etc. in scientific statements (knowledge claims) or suggesting new, maybeinterdisciplinarylinesofresearchbasedonsurprisingrelationshipsbetweendata.

Recommendations

o LODbasedapplicationsthatenableadvancesinarchaeologicalresearchprocessesandoutcomesmayfosteruptakeoftheLODapproachbytheresearchcommunity.

o LODbasedapplicationsforresearchwillhavetodemonstrateadvantagesoverorotherbenefitsthanalreadyestablishedformsofdataintegrationandexploitation.

o DevelopLODbasedservicesthatgobeyondsemanticsearchandretrievalofinformationandalsosupportotherresearchpurposes.

o BuildontheCIDOCCRMandavailableextensionstoexploitconceptuallyintegratedLOD.



7 LinkedDatadevelopmentinARIADNE

TheARIADNEprojectpromotesacultureofopensharingand(re-)useofarchaeologicaldataacrossinstitutional,nationalanddisciplinaryboundariesofarchaeologicalresearch.LinkedOpenDatacangreatlycontributetothisgoal.ThereforeARIADNErecognisesLinkedDataasakeyapproachfordatasharingandinteroperability.Onestrandoftheprojectworksupportsthedevelopmentofsuchdata.Theactivitiesinthisstrandofworkconcerned

o themetadataofthedatasetsregisteredintheARIADNEdatacatalogue,

o vocabularies for the metadata describing registered datasets (e.g. mapping of existingvocabularies,supportforthegenerationofvocabulariesinSKOS),

o mappingofdatasetstothecoreCIDOCCRMandextensionsoftheCRMcreatedinARIADNE,

o demonstrators generating and using Linked Data (e.g. metadata extracted from unstructureddatasuchasgreyliterature,CIDOCCRMbaseddatasets),and

o providingaccesstoARIADNELinkedDataforexternalapplicationdevelopers.

ThustheworkmainlycentredonLinkedDatarelatedtodataregistration,enablingdataintegrationviavocabulariesandtheCIDOCCRMontology,demonstrationofenhancedornewcapabilities(e.g.enhancedcross-searchingofdataresources),andpreparingthegroundforlinkingofresourcesalsobeyond the ARIADNE pool of resources. The ARIADNE data catalogue and other results of theactivitieslistedaboveareincludedintheARIADNEgraphdatabaseandaccessiblethroughaSPARQLendpoint (seeChapter8).Thesectionsbelowdescribe theactivities ingreaterdetail, including theLinkedDatamethodsandtoolsthathavebeenapplied,enhancedornewlydevelopedbyARIADNEresearchersanddevelopers.

7.1 TheARIADNEcatalogueasLinkedOpenData

ThekeycomponentoftheARIADNEe-infrastructureisthedatasetregistry/catalogue.Intheregistrydataprovidersdescribetheirresources(datasets,collections,etc.)basedonacommonmodel,theARIADNECatalogueDataModel(ACDM)197.TheACDMbuildsontheW3C’sDataCatalogVocabulary(DCAT)198whichhasbeendesignedtofacilitateinteroperabilitybetweendatacatalogspublishedontheWeb.TheACDMextendsDCATtakingaccountofrequirementsofdescribingarchaeologicaldataresources.TheARIADNEregistry/catalogueholdsmetadataofdataresources, theprojectdoesnotcollect, store and curate primary research data – which are tasks of the data providers (e.g.communitydataarchivesorinstitutionalrepositories).Themetadataisbeingcollectedandenrichedwith the MoRe (Metadata & Object Repository) aggregator199 and included in the ARIADNE datacatalogue.ARIADNEmakesthecatalogueandotherdatageneratedintheprojectavailableasLinkedOpen Data. This means that other service/application developers can query the data as well asinterlinkitwithotherLOD.TherebytheARIADNELODcanbecomepartofaLinkedData“cloud”ofarchaeologicalandrelatedotherinformationresources.

197ARIADNECatalogueDataModel(ACDM),http://support.ariadne-infrastructure.eu198W3C(2014)Recommendation:DCAT-DataCatalogVocabulary,16January2014,

http://www.w3.org/TR/vocab-dcat/199MoRe(Metadata&ObjectRepository),http://more.dcu.gr;alsoregistrationofsingledatasetswiththe

metadataenteredmanuallyispossible.



7.2 WorkonvocabulariesasLinkedData

ProjectpartnersconductedvariousworkconcerningvocabulariesasLinkedData.Thisincludes

o GenerationofSKOSversionsofexistingornewlydevelopedvocabularies,

o Development of a toolset for vocabularymapping andmapping of subject vocabularieswhichpartnersusefordataindexingtoamajorcommonvocabulary,theArt&ArchitectureThesaurus,

o Use of vocabularies to support Natural Language Processing (e.g. metadata extraction fromarchaeological“greyliterature”,

o MappingofdatasetstothecoreCIDOCCRMandextensionsoftheCRMcreatedinARIADNE,

o DemonstratorsusingLinkedData(e.g.CIDOCCRMbaseddatasets)anddemonstratingenhancedornewcapabilities(e.g.enhancedcross-searchingofdataresources).

Thisworkandresultsachievedaredescribedinthesectionsthatfollow.

7.2.1 VocabulariesinSKOS

Vocabulariessuchastaxonomiesandthesauriareessentialknowledgestructuresandterminologyofdomainsofknowledge.ARIADNEisaprojectandthereforenotinapositiontopublishandmaintainvocabularies. This must be done by the institutions who own the vocabularies. However somepartners and associated organisations own and/or manage national or other major vocabularies,which are being used in ARIADNE. Below we briefly describe vocabularies that have beentransformedtoSKOSpreviously, inparalleltoorwithintheARIADNEproject, includingthenumberofmappingstotheArt&ArchitectureThesaurus(whichisdescribedinthenextsection):

o ItalianMinistryofCulturalAssetsandActivities/CentralInstitutefortheUnionCatalogue(ICCU)– PICO thesaurus200: A large thesaurus related to culture and cultural heritage (Italian andEnglish)which isbeingusedforthedataofCulturaItalia201;asmallnumberofabout200termsconcernarchaeologyofwhichmosthavebeenmappedtotheAAT.

o GermanArchaeological Institute (DAI) vocabularies: The Institutehasvocabularies fordifferententities (e.g. books, collections, inscriptions, buildings and structures, multi-part monuments,topographicobjects)fromwhichabout400concepts,alreadyinSKOSandpreviouslymappedtotheAAT,arebeingusedinARIADNE.WorkisongoingtoharmonizethedifferentDAIthesauritoonecommonstandard,theiDAI.vocab202.

o MajorUKthesauri203: IntheSENESCHALproject(UK,AHRC-fundedproject,2013-2014),runningin parallel to ARIADNE, the project partner University of South Wales (Hypermedia ResearchGroup)helpedUKheritageinstitutions–HistoricEnglandandtheRoyalCommissionsonAncient&HistoricalMonumentsofScotland (RCAHMS)andWales (RCAHMW)make their vocabularies

200PICOthesaurus(MiBAC-ICCU,Italy),http://purl.org/pico/thesaurus_4.2.0.skos.xml201CulturaItalia:Dati,http://dati.culturaitalia.it202iDAI.vocab:Thisisagroupof14thesauriofmonolingualarchaeologicalterminologyaimedtocollectand

organisetheterminologyusedininformationservicesoftheGermanArchaeologicalInstitute.Thethesauriareindifferentlanguages(Arabic,Chinese,English,Farsi,French,German,Greek,Hungarian,Italian,Portuguese,Russian,Spanish,Turkish,Ukrainian)andofvariedsize(rangingfrombelow100toseveralthousandterms).TheGermanthesaurus,whichisalreadymappedtotheAAT,servesasthecentralhubtoandthroughwhichtheotherthesauriarelinked.iDAI.vocab,http://archwort.dainst.org

203HeritageData-LinkedDataVocabulariesforCulturalHeritage,http://www.heritagedata.org



availableinSKOSformatasLinkedOpenData.InARIADNEtheArchaeologyDataServiceemploysfiveHistoricEnglandthesauriofwhichabout850conceptshavebeenmappedtotheAAT.

o Fédérationetressourcessurl’Antiquité(FRANTIQ,France)–PACTOLSthesaurus204:Alargemulti-lingual thesauruswhich focusesonantiquityandarchaeology fromprehistory to the industrialage;termsinFrench,English,German,Italian,Spanish,Dutch,and(some)Arabic).ARIADNEhasacooperationagreementwithFRANTIQonthedeploymentofPACTOLSintheproject.Over1600PACTOLSconceptswhichtheARIADNEpartnerInstitutNationaldesRecherchesArchéologiquesPréventives (Inrap,France)uses in theircatalogueofarchaeological reports (DOLIA)havebeenmappedtotheAAT.

o IntheNetherlands,DataArchivingandNetworkedServices (DANS)providea listofmonumenttypes(Archeologischecomplextypen)fordescribingDutcharchaeologicalexcavations.Thetypesaremanaged by the Rijksdienst voor het Cultureel Erfgoed (RCE)205. These have recently beenexpressedasSKOS.About450conceptshavebeenmappedtotheAAT.

o The most detailed classification system available for Irish Monument types is the class listdevelopedbytheNationalMonumentsService(NMS).Thisisahierarchicallistwhichwasusedinthe classification of sites and monuments that formed part of the Archaeological Survey ofIreland.IthasbeenexpressedinSKOSaspartoftheLoCloudproject206.Over480conceptshavebeenmappedtotheAAT.

o AIAC’sFASTIOnlineusesa flat listofmonumenttypes in the“advanced”search interface.ThesetofFASTIconceptsarepublishedonlinewithURIs207.About130conceptshavebeenmappedtotheAAT.

Within the ARIADNE project data providers, with support by the University of South Wales(Hypermedia Research Group), created or transformed/enhanced existing vocabularies in/to SKOSformat:

o Data Archiving and Networked Services (DANS, Netherlands) – Dendrochronologymulti-lingualvocabulary:WithhelpfromARIADNE,DANSandcollaboratorshaverestructuredandenhancedtheTreeRingData Standard (TRiDaS). TRiDaS208 is used todescribe thedata resulting fromallkinds of dendrochronological analysis. The multilingual vocabulary, which has recently beenexpressed in SKOS, is being employed for the Digital Collaboratory for Cultural Dendro-chronology209 (Jansma 2013) and available also to other users. Some 336 concepts have beenmappedtotheAAT.

o ItalianMinistryofCulturalAssetsandActivities/CentralInstitutefortheUnionCatalogue(ICCU)– Reperti Archeologici (RA) Thesaurus210: A pictorial thesaurus describing archaeological finds.This has been expressed as SKOS during ARIADNE using the STELLAR toolkit. About 1100conceptsofthisvocabularyhavebeenmappedtotheAAT.

204PACTOLS(Peuples,Anthroponymes,Chronologie,Toponymes,Œuvres,LieuxetSujets),

http://pactols.frantiq.fr205See:http://cultureelerfgoed.nl/dossiers/archis-30/archeologisch-basisregister-plus206IrishMonumentshttp://vocabulary.locloud.eu/Irish_Monuments/207FASTIOnline,seehttp://www.fastionline.org/data_view.php,andforanexampleofaconceptwithURIsee

http://www.fastionline.org/concept/attributetype/monument208TRiDaS-TheTreeRingDataStandard,http://www.tridas.org209DigitalCollaboratoryforCulturalDendrochronology-DCCD,http://dendro.dans.knaw.nl;projectwebsite:

http://vkc.library.uu.nl/vkc/dendrochronology/210RepertiArcheologici(RA)Thesaurus,http://www.iccd.beniculturali.it/index.php?it/473/standard-

catalografici/Standard/74;http://vast-lab.org/thesaurus/ra/vocab/index.php



7.2.2 Mappingofsubjectvocabularies

Themain goal of themapping between vocabularies in the ARIADNE project has been to enablesearchingofrelevantdataresourceswhicharebeingheldbyarchivesindifferentcountries.Bringingtogethertheoriginalresourcemetadatadoesnotallowforeffectivesearchingofrelevantresources,becausetheprovidersusetermsfromsubjectvocabulariesindifferentlanguagesand,ifinthesamelanguage,oftenusedifferenttermsforthesamesubject.

To enable cross-searching of data resources mapping of terms was necessary. But the ARIADNEprojecthas15dataprovidersandmanyothersexpressedinteresttomakedataresourcessearchablethrough the ARIADNE portal. There is no scalable approach for direct, many-to-many mappingbetween terms in several vocabularies. Therefore it was decided to use an appropriate commonvocabularyas intermediary“hub”ontowhichdataprovidersmaptheirsubjectterms(thesocalledswitchinglanguageapproach).Thecontent-richandmulti-lingualArt&ArchitectureThesaurus(AAT)of the Getty Research Institute has been selected as the central hub of themapping. The AAT isavailableasLinkedOpenDatainSKOS,publisheduntertheOpenDataCommonsAttributionLicense(ODC-By)1.0211.

TheAATcontainsover40,000conceptsandover350,000terms,organised insevenfacets (and33hierarchies as subdivisions): Associated concepts, Physical attributes, Styles and periods, Agents,Activities,Materials,Objectsandoptionalfacetsfortimeandplace(Harpring2016).TheAAT’sscopeis broader than archaeology, encompassing visual art, architecture, other material heritage,archaeology, conservation, archival materials, etc., but contains many useful high levelarchaeologicalconcepts,particularlyintheBuiltEnvironment,MaterialsandObjectshierarchies.

Vocabularymappingtools

For the mapping the project partner University of South Wales (Hypermedia Research Group)developedaninteractivetoolwhichenablessubjectexpertstoproduceSKOSmappingrelationships(e.g. broadMatch or closeMatch) between their vocabulary terms and the AAT terms (Binding &Tudhope 2016). The tool is a lightweight browser based application that presents concepts fromchosensourceandtargetvocabulariessidebyside,exposingadditionalcontextualevidencetoallowthe user to make a more informed choice when deciding on potential mappings. The tool is forvocabulariesalreadyexpressedinRDF/SKOSandcanworkdirectlywiththedata–queryingexternalSPARQLendpointsratherthanstoringanylocalcopiesofcompletevocabularies.Thesetofmappingsdeveloped can be saved locally, reloaded and exported to a number of different output formats(JSONforuse inARIADNE).Thetool isprovidedopensourceandthesoftwarecode isavailableonGitHub212.Asecondmappingapproachhasbeendevelopedforsourcevocabulariesthataresmallerterm lists and not yet expressed in RDF. Such term lists are often available or can be easilyrepresentedinaspreadsheet.Astandardtemplatewithexamplemappingswasdesignedtosupportdomainexpertsinthemappingoftermstothetargetvocabulary.ACSVtransformationproducestherepresentationofthemappingsinRDF/JSONformat213.

211GettyVocabulariesasLinkedOpenData,http://www.getty.edu/research/tools/vocabularies/lod/index.html212VocabularyMatchingTool,http://heritagedata.org/vocabularyMatchingTool/;sourcecodeforlocal

downloadandinstallation,https://github.com/cbinding/VocabularyMatchingTool213ARIADNEsubjectmappings:Spreadsheettemplateandconversion,https://github.com/cbinding/ARIADNE-

subject-mappings



Mappingsconducted

The application of the tools and the “hub” approach have first been tested and evaluated in anexploratory pilot (Binding & Tudhope 2016). Terms of five subject vocabularies employed byARIADNE data providers were mapped to the AAT and the semantic linkage used for retrievalexperiments. The vocabularies are: a flat list of monument types employed in Fasti Online (inEnglish), terminology for types of archaeological sites of the Central Institute for the UnionCatalogue, Italy (in Italian), Archeologische complextypen of the Rijksdienst Cultureel Erfgoed (inDutch, employed by Data Archiving and Networked Services, Netherlands), relevant terms of thearchaeologicaldictionaryoftheGermanArchaeologicalInstitute(inGerman),andHistoricEngland’sThesaurus of Monument Types (in English, employed by the Archaeology Data Service, UK). Thestudy demonstrated advantages of the approach by performing mediated cross-search overarchaeological datasets from different countries with semantic expansion across the multilingualvocabularies.

ByJune2016,conceptsfrom25vocabulariesemployedby11projectpartnerswerealreadymappedto the AAT; six partners each employed concepts from 1 vocabulary, two partners each from 2vocabularies,and theother threepartners from4,5and6vocabularies. In termsof structureandsize the vocabularies varied from a small term list for a particular dataset to standard nationalvocabularieswithalargenumberofconcepts.15ofthevocabularymappingswereconductedwiththespreadsheettemplate(orasimilarpartnerspreadsheet),2usingtheonlineinteractivemappingtool (i.e. when the source vocabulary was available in RDF/SKOS) and 8 using the partner’s own(intellectual/manual)resources.

Intotal5823mappingswereconducted,withmappingsofindividualpartnersrangingfromafewuptoover1600terms.Togivesomeexamples:TheInstituteofArchaeologyoftheScientificResearchCentreoftheSlovenianAcademyofSciencesandArts(Slovenia)mapped93termsforarchaeologicalsiterecordsintheirARKAS-ArheološkikatasterSlovenijesystemtotheAAT;theDataArchivingandNetworkedServices(Netherlands)andcollaboratorsmapped336conceptsofthevocabularyoftheDigital Collaboratory for Cultural Dendrochronology, the Discovery Programme (Ireland) 486concepts of the Irish Monument Types thesaurus, the Institut National des RecherchesArchéologiquesPréventives(France)1634conceptsofthePACTOLSthesauruswhicharebeingusedbytheircatalogueofarchaeologicalreports(DOLIA).

Very few terms could not be mapped to the AAT. 50% of the mapping relations were skos:exactMatch, 18% skos:closeMatch, 27% skos:broadMatch and 5% skos:narrowMatch (one partneralso did a few skos:relatedMatch mappings). As expected there was only a small number ofskos:narrowMatch mappings, i.e. where the ATT was more specialised than the partners’vocabularies.AnARIADNEproject deliverable is availablewhichdescribes themappings in greaterdetail(ARIADNE2016b).

The ARIADNE data catalogue employs theMoRe (Metadata & Object Repository) aggregator214 toharvestthemetadataprovidedbytheprojectpartnersutilisingtheOpenArchivesInitiativeProtocolforMetadataHarvesting(OAI-PMH).AbespokeAATsubjectenrichmentservicehasbeendevelopedthatappliesthepartnervocabularymappings(inJSONformat)tothepartnersubjectmetadataandderivesanAATconcept(bothpreferredlabelandURI)toaugmentthesubjectmetadatainthedatacatalogue. For example, 773,600 of the Archaeology Data Service or 6131 records of Fasti Onlinehavebeenenriched in thisway.Thecataloguemetadata is supplied to theARIADNEportal,wherethesearchfunctionalitycanusetheAATbasedterminology“hub”toretrievemetadataofdifferent

214MoRe(Metadata&ObjectRepository)aggregator,http://more.dcu.gr



dataproviderswhomappedrelatedsubjecttermstotheAAT.Asearchonatermoriginatingfromany one vocabulary can utilize the mediating structure to route through to terms from othervocabularies (which may be expressed in different languages) and retrieve the identified datarecords.

7.2.3 MetadataforvocabulariesandmappingsinSKOS

Concerning the vocabularies and mappings between them in Linked Data format it would bebeneficialhavingmetadatafortheseproducts.IntheSENESCHALprojectUniversityofSouthWales(HypermediaResearchUnit)producedVoID(VocabularyofInterlinkedDatasets)215metadataofeachof theUK thesauriwhich have been transformed to LinkedData in RDF/SKOS. Thismetadata andlinks to example resources have been published in the DataHub216. Also datasets of mappingsbetweenvocabulariesarevaluablesemanticassetsforwhichmetadataaboutversions,authorship,licensing, etc. would be necessary for users and machines, for example to distinguish betweendifferent mappings produced for large vocabularies. ARIADNE partners who own vocabularies inSKOSandhaveproducedmappingstotheAAThavebeenrecommendedtofollowthegoodpracticeexemplifiedbyUniversityofSouthWales(HypermediaResearchGroup).

7.3 What–Where–WhenasLinkedData

OntheARIADNEdataportalthecoreservicesforcross-searchingthedifferentresourcesforrelevantinformationarebasedonthe“What-When-Where”approach.TheapproachhasbeensuccessfullydemonstratedintheARENAportalforsearchingarchaeologicalsitesandmonumentsofsixEuropeancountries217. In a nutshell, “What” concerns the subjects, “Where” the geographical locations, and“When”theperiods(namedculturalperiodsanddateranges)forwhichuserswishtofindrelevantdata. This information is provided by the data providers in the metadata of the resources theyregisterintheARIADNEcatalogue.

The ARIADNE data portal allows searching across the various data resources based on subjects,location and date ranges (chronology). In the portal this has been implemented as subject-basedsearch, map-based search and a timeline feature. The implementation of the search & browseservices is not based on LinkedData, but such data for subjects, location and chronology is beingprepared, particularly for future linking to external Linked Data resources as well as externaldeveloperswhowishtoquerytheARIADNELinkedDataand/orlinkitwithotherdata.

7.3.1 What(subjects)

LinkedDataforthesubjectscontainedinthemetadatapartnershaveprovidedtotheARIADNEdatacataloguehasbeenproducedthroughthemappingofconceptstotheArt&ArchitectureThesaurus(asdescribedinthesectionsabove.

215W3C(2011)InterestGroupNote:DescribingLinkedDatasetswiththeVoIDVocabulary,3March2011,

http://www.w3.org/TR/void/216HeritageDataonDataHub,http://datahub.io/dataset?q=heritagedata217ARENA-ArchaeologicalRecordsofEurope-NetworkedAccessproject(2001-2004,and2009-2010inthe

contextofDARIAH),http://ads.ahds.ac.uk/arena/search/



7.3.2 Where(places)

“Where”concernsgeographicinformationwhichcanmeanjustnamesofplaces,areas,regions,etc.,or names together with geo-referencing (lat./long coordinates). In the ARIADNE survey onexpectationsfordataportalservicesmap-basedsearchwasaclear“musthave”(cf.ARIADNE2015e:278-289). Therefore the dataset metadata in the ARIADNE catalogue in addition to place namesshould includestandard lat./long.coordinates toallowformap-basedsearchof relevant resourceson the data portal. As the common standard ARIADNE adoptedWGS84 (World Geodetic System1984)218.Most data providers already hadWGS84 based coordinates. In cases where the originalmetadata contained only place names the data providers employed the GeoNames gazetteer toderivecoordinatesforthenames.

ThedatabaseoftheGeoNames219gazetteerisintegratinggeographicaldatasuchasnamesofplacesinvariouslanguages,elevation,populationandothersfromvarioussources.Alllat./long.coordinatesareinWGS84(WorldGeodeticSystem1984).TheGeoNamesdataisavailablethroughanumberofweb services and a daily database export. The data is provided free of charge under a CreativeCommonsAttributionlicense(CC-BY).Itcontainsover10milliongeographicalnamesandconsistsofover9millionuniquefeatureswhereof2.8millionpopulatedplacesand5.5millionalternatenames.

GeoNames is available as Linked Open Data and one of the core linking hubs of the Linked DataCloud.ThereforeARIADNEseesGeoNamesasthecoregazetteerforLinkedDatabasedlinkingwithexternal data resources based on place names and other geographical information. GeoNamescovers modern places and other geographical information, which is also generally used byarchaeologistsinthedocumentationoffieldwork,reportsandpublications.Howeverarchaeologicalmaterial alsooften includes ancient/historical placenames andother geographical references. Forsuch references ARIADNE itends to collaborate with the Pelagios initiative which employs thePleidades and other Ancient World gazetteers. The ARIADNE partners German ArchaeologicalInstituteandFastiOnlinealreadyparticipateinthePelagiosproject(seeSection5.3).

7.3.3 When(chronology)

In archaeology the “when” of sites and objects is typically given as a cultural periods and date-ranges. In the ARIADNE survey on expectations for the data portal services the archaeologicalresearchers considered searching data resources based on cultural periods and date-ranges asparticularlyimportant(cf.ARIADNE2015e:278-289).

Toenablesuchsearching,datapartnershavetogiveintheirmetadatatheperiodtermswhichtheyuse and the absolute date ranges (start/end dates) which apply to each term for theircountry/regions. The period terms and date ranges are often defined in standard nationalperiodizations but also proprietary controlled period lists derived from authoritative sources arepossible.Forexample, theArcheologischBasisregister (ABR)of theCulturalHeritageAgencyof theNetherlandsorMIDASHeritagefortheUKprovidestandardnationalperiodizations.

A cultural period as elaborated in archaeological and historical research has temporal andgeographicalboundaries,definedby somecharacteristicswhich set it apart fromthepreviousandlaterperiodinachronology.NamedperiodsearchontheARIADNEdataportal,forexample“Roman”returns results for periodAD43 toAD410 fromUKdatasets and results for period 10BC toAD450

218WorldGeodeticSystem1984(WGS84),http://earth-info.nga.mil/GandG/wgs84/219GeoNames,http://www.geonames.org



fromDutchdatasets;howeverdate-range/timeline-basedsearch,e.g.10BC toAD40 returnRomanresultsfromDutchdatasetsandIronAgeresultsfromUKdatasets.

On Linked Data for cultural periods ARIADNE collaborates with the PeriodO project220. PeriodO isbuildingasystemforcollecting,organisingandreferencingdefinitionsofperiodsbasedonURIs.TheperiodsareprovidedthroughanonlineapplicationaswellasadownloadablesetofLinkedData.ThePeriodOapproach is togather individualperiodassertionsmadebyauthoritative scholarly sourcesabout thetemporalandspatialboundariesofperiods inparticular researchcontexts, retaining theprovenanceof theassertions,e.g. scholarlybookorpaper (Rabinowitz2014;Golden&Shaw2015and2016).

But the PeriodO system also includes established national periodizations. ARIADNE has producedfromavailableperiodizationsasetofculturalperiodsandtheir timeranges fromthePaleolithic toModern times for 24 European countries (in total 659 periods)221. The periods set has beenincorporatedinthePeriodOsystemwhichallowsstablelinkingofdatabasedonthepersistentURIsassignedbyPeriodO.TousethePeriodOURIsinARIADNEanenrichmentserviceisbeingdevelopedand included in the MoRe aggregator which will attach the URIs when processing the metadataharvestedfromdataproviders.

ThroughthePeriodOsystemalsootherprojectscanuseperiodsprovidedbyARIADNEandothers.ARIADNE promotes the use of PeriodO URIs to allow for wider interlinking of data based onperiods/chronologies. The PeriodO project is funded until 2018 by a grant of the US Institute ofMuseumandLibraryServices.

7.4 UseofvocabulariesinNLPanddatamining

Vocabulariesarealso important innatural languageprocessinganddataminingtasks.ThesectionsbelowdescribesuchusesinresearchanddevelopmentcarriedoutinARIADNE.

7.4.1 NaturalLanguageProcessing

InARIADNEalsoresearchanddevelopmentonNaturalLanguageProcessing(NLP)ofarchaeologicalcontent has been explored with the aim of making text-based resources more discoverable anduseful (ARIADNE 2015c). This work of researchers of the Archaeology Data Service, University ofSouthWales (HypermediaResearchGroup)andLeidenUniversity (FacultyofArchaeology) focusedspecificallyonthe“greyliterature”ofarchaeologicalinvestigations.

Thepartnershaveexploredmachinelearningandrule-basedapproaches.Herewefocusontheworkonruled-basedmethods inwhichvocabularies inLinkedData formathavebeenused. In thisworktheOPTIMAsemanticannotationsystemoftheHypermediaResearchGrouphasbeenused.OPTIMAperforms theNLP tasks ofNamed Entity Recognition, Relation Extraction,NegationDetection andWord-SenseDisambiguationusinghand-crafted rulesand terminological resources (Vlachidis2012;Vlachidisetal.2013;Vlachidis&Tudhope2015a).ThesystemusestheGATE(GeneralArchitecturefor Text Engineering) framework, Ontology Based Information Extraction (OBIE) and several othertechniques.

OPTIMA contributed to the Semantic Technologies for Archaeological Research (STAR) project, apioneerintheuseofNLPforextractionofmetadataandlinkingofarchaeologicalgreyliteratureand220PeriodO-Periods,Organized,http://perio.do;seealsohttps://wiki.digitalclassicist.org/PeriodO221ARIADNEsetofculturalperiodsinthePeriodOsystem,http://n2t.net/ark:/99152/p0qhb66



digital archive databases based on English Heritage terminology vocabularies and the CIDOC CRM(Tudhopeetal.2011b;Vlachidisetal.2012).

TheNLPwork inARIADNEbuildsupontheexperiencesofSTARbuttargets“grey literature”also inotherlanguages.Thisfaceschallengesofdifferentvocabularies(e.g.withregardtostructure)aswellasdifferencesinlanguagecharacteristics.TheaddressthesechallengesgreyliteratureinDutchhasbeenchosenusingthesaurioftheRijksdienstCultureelErfgoed.TheoriginalSKOSifiedthesauriwerenot suitable for supporting Ontology Based Information Extraction (OBIE) approaches, due to theincapacity of the GATE ontology tool to parse (understand) broader/narrower term relationships.ThereforetransformationofthethesauritoOWL-Lite(ontology)wasnecessary.

With regard to language characteristicsparticularly compoundnoun formspresenta challenge forthe usual “whole word” matching mechanisms. Compound noun forms examples might include“beslagplaat”whereboth “beslag” and “plaat” are known to the vocabulary andalso “aardewerk-magering”whereaardewerk(pottery)isknownbut“magering”isnot.

But the current pilot system has achieved some promising semantic enrichment of Dutch greyliterature reports, concerning artefacts (such as “aardewerk”) and other concepts including timeperiods.Inordertoovercomethe“wholeword”restrictionsmechanismsoperatingonpartmatchingare being explored. Negation detection is another aspect that has been explored during ARIADNE(Vlachidis et al. 2015b); it is important to distinguish whether the text indicates that evidence ofsome archaeological issue has or has not been found during an excavation. Expansion of NLP forextraction, indexing and linking of data/metadata fromother European language grey literature isintended. Critical for good results in general is the availability of rich and well-structuredvocabularies,buteveninsuchcasessomemodificationmayberequiredtoconductNLPwithoptimalresults.

7.4.2 MiningofLinkedData

ARIADNE partner Leiden University, in collaboration with the associated partner Free UniversityAmsterdam, examined the feasibility ofmining archaeological LinkedData, for example, to detectrelevantpatternsinthegraph-structureofsuchdata.

In the first years of the project, started in February 2013, no archaeological Linked Data wasproducedintheproject.Butanexaminationofafewdatasetsavailableelsewhereshowedthattheylargelyconsistedofflatdatastructureswithdescriptivemetadatavalues(ARIADNE2015b).Miningofsuchdata is unlikely to yield archaeologically interestingpatterns. Indeed, interviewswithdomainexperts indicated a strong interest in archaeological contexts, which means rich informationgenerated in fieldwork. Particularly interesting would be spatio-temporal patterns betweenarchaeologicalcontexts.

Thereforetheresearchgroupdecidedtoworkon information intheDutcharchaeologicalprotocolSIKB 0102, called digital “pakbon” (package slip), developed and maintained by the StichtingInfrastructuur Kwaliteitsborging Bodembeheer (SIKB) / Foundation Infrastructure for QualityAssuranceofSoilManagement222.TheSIKB0102hasbeenintroducedafewyearsago(firstversionin2010).Itspecifieswhichmandatoryinformationaboutexcavationsandfindshastobeprovidedasan XML document when depositing data in the E-Depot for Dutch Archaeology (managed by

222StichtingInfrastructuurKwaliteitsborgingBodembeheer:Protocol0102Archeologie,

http://sikb.nl/datastandaarden/richtlijnen/protocol-0102



ARIADNEpartnerDataArchivingandNetworkedServices-DANS)223.Withregardtoterminologythethesauri in the Archeologisch Basisregister (ABR+) of the Rijksdienst Cultureel Erfgoed (CulturalHeritageAgency)224havetobeused.

While the amount of “pakbonnen” is growing each one still is an isolated entity and the XMLdocumentsassuchcannotbeusedforsemanticintegrationandminingoftheinformation.Thereforethe research group developed a Linked Data version of the SIKB 0102 (pakbon-ld), whichincorporates its set of archaeological concepts and properties, but restructured and expanded toexploit thegraph structure225. This versionhasbeenmodelled inCIDOCCRM including theEnglishHeritageextension(CRM-EH)whichcontainsarchaeology-specificconceptsandrelations.MoreoverABR+ thesauri in SKOS have been prepared for use in the transformation of SIKB 0102 XMLdocuments to Pakbon LinkedData.Once these foundationswere completed, a tool for automaticconversionhasbeendeveloped226.Withthistool73SIKB0102XMLdocumentsfromtheE-DepotforDutchArchaeologyhavebeentranslatedandstoredinthegraphdatabasetogetherwiththeCIDOCCRM,CRM-EHandABR+vocabularies.

So far the results of mining this resource with SPARQL queries have been encouraging from atechnical point of view, but far from useful from an archaeological perspective (e.g. trivial orconflictingresults).Itappearsthatthedetectionofarchaeologicallymeaningfulpatternsrequiresaniterative interactionof researcherswithquery results fromadatabaseof still richerdata than the“pakbonnen”provide.But theproject nowhas amodel and tool for convertingdocumentationoffieldworkintheNetherlandstoLinkedDataandincludeitinthewebofarchaeologicalLinkedData.

223E-depotforDutchArchaeology,http://www.edna.nl224RijksdienstCultureelErfgoed:ArcheologischBasisregister,http://abr.erfgoedthesaurus.nl225WilkeXander(VUAmsterdam,SPINlab):PakbonLinkedData,http://pakbon-ld.spider.d2s.labs.vu.nl/home226WilkeXander(VUAmsterdam,SPINlab):LinkedDatatranslationoftheSIKBarchaeologicalprotocol0102

(akaPakbon),https://github.com/wxwilcke/pakbon-ld



7.5 CIDOCCRMextensionsandmappings

ARIADNErecommendstheCIDOCConceptualReferenceModel(CRM)227asacommonontologyfordata integration,discoveryandaccessbasedonLinkedData, including themoreambitiousgoal tosupportresearch-orientedapplications(seeSection6.6.5).

The CIDOC CRM has been developed specifically for describing and facilitating the exchange andintegrationofculturalheritageknowledgeanddata.Archaeologypartlyoverlapswiththisdomainaswellasneedsmodellingofadditionalconceptualknowledge,forexample,todescribeobservationsofanexcavation(e.g.stratigraphy).TheARIADNEReferenceModelcomprisesthecoreCIDOCCRMand a set of enhanced and new extensions, including the archaeological excavation process(CRMarchaeo)andbuiltstructuressuchashistoricbuildings(CRMba).

ThetablebelowgivesanoverviewoftheextensionstotheCIDOCCRMwhichhavebeencreatedorenhancedintheARIADNE228:

o CRMgeo:spatio-temporalmodelthatarticulatesrelationsbetweenthestandardsofthegeospatialandtheculturalheritagecommunities(integratesCRMwithOGCstandards;applicationssuchasGeoSPARQL)

Newextension,v1.0,April2013

o CRMdig:modelofdigitisationprocesses,toencodemetadataaboutthestepsandmethodsofproduction(“provenance”)ofdigitalrepresentationssuchas2D,3Doranimatedmodels(validatedinseveralprojects)

Enhancedextension,v3.2,August2014

227CIDOC-ConceptualReferenceModel(CIDOC-CRM),http://www.cidoc-crm.org228DescriptionoftheARIADNEReferenceModelandindividualextensions(includingreferencedocument,

presentation,RDFSencoding)isavailableathttp://www.ariadne-infrastructure.eu/Resources/Ariadne-Reference-Model;seealsohttp://www.ics.forth.gr/isl/CRMext/



o CRMsci:modelforintegratingmetadataaboutscientificobservation,measurementsandprocesseddata(validatedinarchaeology,biodiversityandgeologycases)

Enhancedextension,v1.2.2,August2014

o CRMinf:modelforintegratingdatawithscholarlyargumentationandinferencemakingindescriptiveandempiricalsciences(beingvalidatedwithscholarlyannotations);harmonizedwithCRMsci

Newextension,v0.7,February2015

o CRMarchaeo:modelforintegratingmetadataaboutthearchaeologicalexcavationprocess(introducesconceptsofstratigraphyandexcavation);beingvalidatedbyarchaeologicalrecords


o CRMba:modelforinvestigatinghistoricandprehistoricbuildings,therelationsbetweenbuildingcomponents,functionalspaces,topologicalrelationsandconstructionphasesthroughtimeandspace;harmonizedwithCRMarchaeo


o ARIADNEReferenceModel:CIDOCCRM+setofneworenhancedextensions

ARIADNEReferenceModel,v1.0,April2016

TheARIADNEReferenceModelisintendedtoallowtheaccuratedocumentationofcomplexentitiesand relations of archaeological/scientific observations and analysis, data integration and search,involvingreasoningoverthedistributeddataandknowledge.Thishoweverdependsontheinterestofdataproviderstomaptheirdatabasestorelevantpartsoftheconceptualreferencemodel,whichsomeARIADNEpartnershavealreadydoneandothersareconsidering(ARIADNE2016a).

CRMmappingtool

A new tool, the Mapping Memory Manager (3M)229 has been developed by ARIADNE partnerFoundationforResearchandTechnologyHellas, InstituteofComputerScience(FORTH-ICS,Greece)to facilitate the mapping of databases to the extended CIDOC CRM and the validation of themapping;mappings canbeexported inCRMcompliantRDF.Themappingprocess is supportedbytheX3MLMappingFrameworkthatensurestheintegrityandpreservationofthe“meaning”oftheinitialdata(Minadakisetal.2016).

Mappingofdatabases

Severalpartnerdatabases(DBschemas)havebeenmappedwiththe3Mtooltorelevantpartsoftheextended CIDOC CRM. Some of the mappings have been used in pilot applications whichdemonstrateadvantagesoftheextendedCRM(seebelow).Thefollowingthreeexamples illustraterepresentativemappings:

dFMRÖ - Digitale Fundmünzen der Römischen Zeit in Österreich (Digital Coin-finds of the RomanPeriod inAustria)230:ThedFMRÖisarelationaldatabaseofpre-RomanandRomanImperialperiodcoins found inAustria andRomania (75,565 records of coin finds), developedby theNumismaticsResearch Group at the Austrian Academy of Sciences. The database schema of the dFMRÖ wasmapped to CIDOC CRM, using also the CRMdig extension and a specialized extension for coinscoveringtheneedtomapcategoricalinformation(Doerretal.2016).Thedatabaseprovidedagood

229MappingMemoryManager-3M(FORTH-ICS),http://www.ics.forth.gr/isl/3M230dFMRÖ-DigitaleFundmünzenderRömischenZeitinÖsterreich(ÖAWNumismaticResearchGroup),

http://www.oeaw.ac.at/antike/index.php?id=358



exampleformappingofalargeclassofwell-definedtraditionaldatabaseswherethereisaneedtoaddressandseparatebothcategoricalandfactualinformation.Resultshavebeenemployedtogetherwithotherdatasetsinthecoinsdemonstrator.

Athenia Agora excavation database: This database (over 280,000 data items) presented a case ofhighlycontextualizedresearchdata.Themostrelevantpartsofthedatabaseschemaweremappedby a researcher of the German Archaeological Institute to CIDOC CRM, using the extensionsCRMarchaeoandCRMsci.Themappingresultshavebeenusedtogetherwithotherdatasets inthesculpturesdemonstrator.

SITAR - Archaeological Territorial Informative System of Rome231: The SITAR system managesdifferenttypesofdatasetsincludinginformationaboutmonuments,archaeologicalfinds,surveyandconservationwork,archivaldocuments,bibliographicreferencesandothers.AmappingbetweentheSITARdatabaseschemaandtheconceptsofCIDOCCRMandCRMarchaeohasbeencarriedoutbytheARIADNEpartnerItalianMinistryofCulturalAssetsandActivities(CentralInstitutefortheUnionCatalogue) in cooperation with domain experts of the Soprintendenza Speciale per il Colosseo, ilMuseoNazionaleRomanoel’AreaArcheologicadiRoma,andtheDepartmentofComputerScienceoftheUniversityofVerona.

AlsotheACDMmodeloftheARIADNEdataregistry/cataloguehasbeenmappedtotheCIDOCCRMandasetof integratedqueries implemented inordertovalidatetheadequacyof themodels.Thismapping isbeingusedtosupportdata integrationbothatthecatalogueandatthe itemlevel.TheenhancedcapabilityprovidedbytheARIADNEReferenceModelisbeingdemonstratedinitem-levelpilotapplications.

7.6 DemonstratorsusingCRM-basedLinkedData

ThreepilotapplicationsarebeingdevelopedtodemonstratethecapabilityoftheextendedCRMtosupport Linked Data use cases of item-level data integration, discovery and access. Thedemonstratorsconcerndifferentobjects(coins,sculptures,woodenmaterial)andareimplementedbydifferentpartners.ItisplannedtointegratethepilotdemonstratorsintheARIADNEdataportal,includingamenuofexemplarqueriesforportalusers.

Thecoinsdemonstrator

ThepilotapplicationhasbeenledbyFORTH-ICSanddemonstratedtheitem-levelintegrationprocessofinformationaboutcoinsfromfivedatasetsbasedontheextendedCIDOCCRM,Nomismaontology(numismatics vocabularies)232 and Art & Architecture Thesaurus (Felicetti, Gerth et al. 2016). Thedemonstrator employed the core CIDOC CRM, the extension CRMdig and a small coin-specificextensionmodellingcategoricalinformation.Thefollowingdatasetshavebeenusedinthedemonstrator:o dFMRÖ-DigitaleFundmünzenderRömischenZeitinÖsterreich(DigitalCoin-findsoftheRoman

PeriodinAustria),onlineMySQLdatabase(source:NumismaticsResearchGroupattheAustrianAcademyofSciences);

231SITAR-SistemaInformativoTerritorialeArcheologicodiRoma,http://www.archeositarproject.it232Nomismaontology,http://nomisma.org/ontology



o MuseiD-Italia documentation of several coins collections of Italian museums integrated inCulturaItalia (source: ItalianMinistryofCulturalAssetsandActivities-Central InstitutefortheUnionCatalogue);

o A subset of numismatics records (1670) from the Fitzwilliam Museum (Cambridge) databasepreparedintheCOINSproject(COINS-CombatOn-lineIllegalNumismaticSales,2007-2009,seeJarrettetal.2011;COINSwasledbyPIN-VastLab,theCoordinatoroftheARIADNEproject);

o Coins data records (630) from the Soprintendenza Archeologica di Roma (SAR) database –preparedintheCOINSproject;

o Documentationofcoinfinds(517) inthe iDAI.fieldresearchdatabaseofthePergamonproject,with detailed information about the archaeological context (source: German ArchaeologicalInstitute).

o Natural Language Processing techniques were employed by University of South Wales(HypermediaResearchGroup)toextractnumismaticinformationfromasamplesetofsixreportsfrom theADSGrey Literature library todemonstrate thepotentialofNLP fordata integration.TheresultingdatawasexpressedinthesameCIDOCCRM,AATandNomismaformusedforthedatasets. Itwassuccessfully integrated intotheFORTH-ICSdemonstratoranditwasfoundthattheNLPtechniqueshadidentifieditemsfromthereporttextnotexplicitlymentionedinthesiterecordmetadata.

The demonstrator aimed at item-level integration of the diverse coin datasets in an environmentwhereuserscaneffectivelyqueryandreceivecombinedresultscomingfromthedifferentdatasets.Toenablesuchasearchenvironment fourof thedatasetsweremappedwithFORTH-ICS’MappingMemory Manager (3M) to the ARIADNE Reference Model and transformed to RDF format; theMuseiD-Italia datawas already in CIDOC-CRMRDF form, compatiblewith the ARIADNE ReferenceModel. Inadditionmappingof terms indataset records to theArt&ArchitectureThesaurus (AAT)andNomismaontology(bothavailableasLinkedData)wasnecessarytoenableintegratedsearchingofthecoinsdocumentation.ThepilotapplicationemploystheBlazegraphRDFgraphdatabase233andtheuserinterfaceisbasedon the Metaphacts platform234. The platform implements the Fundamental Categories andRelationshipsforintuitivequeryingCIDOCCRMbasedrepositories,describedinTzompanaki&Doerr(2012).Userscanformulatequeriesbyselectingfromsixbasiccategoriesandtherelationsbetweenthemwithout the need to be familiarwith the underlying schema. The results of the queries arecomingfromthedifferentdatasets,anditispossibletorefinethesearchwithafacetview.The coin demonstrator has shown that datasets of different origin, language, property, and ofheterogeneousinformationcanbesuccessfullyintegratedbyrelyingontheCIDOCCRM.Therelativehomogeneityofthecoinclassofobjectshasmadethemappingandconversionworkrelativelyeasy.Butvalidityofthemethodologicalapproachcanbeassumedforanytypeofarchaeologicalobject.

Thesculpturesdemonstrator

ThisdemonstratorhasbeendevelopedbyresearchersoftheGermanArchaeologicalInstitute(Gerthet al. 2016a/b). The researchers produced and explored a dataset of semantic data from fivedifferentdatabasesbasedontheCIDOCCRM,includingtheextensionsCRMsciandCRMarchaeofordescribing scientific data acquisition and archaeological excavation processes. Furthermore thedemonstratorusedtheobject-orientedversionofFunctionalRequirementsforBibliographicRecords

233Blazegraph,https://www.blazegraph.com234Metaphacts,http://www.metaphacts.com



(FRBRoo)235 for describing bibliographical records and the Basic Geo vocabulary236 for simplegeometry description. The researchers developed a prototypical implementation of the differentstandards for archaeological research regarding time, space, actors, literature and other entitiescoveredbydomain-specificvocabulary.Thefollowingdatasetshavebeenusedinthedemonstrator:o GermanArchaeologicalInstitute:Arachne237anddatafromtheiDAI.fieldinstanceoftheChimtou

project238,o BritishMuseum:SemanticWebCollectionOnline239,o OxfordRomanEconomyProject:StoneQuarriesDatabase240,o AmericanSchoolofClassicalStudiesinAthens:AthenianAgoraExcavationdata241.The pilot application presents a case of integration of various datasets with different origins(museumcatalogue,objectdatabase,excavationdatabase,researchresults).Thedataresourcesareprovided with different services and interfaces and therefore required a novel strategy forintegration,basedonCIDOCCRM.ThedataoftheBritishMuseumcouldbeaccesseddirectlyviaitsSPARQLendpointsand integratedbyusingaSPARQLfederatedquery; theBritishMuseumhasthedata already organised based on CIDOC CRM. Arachne’s data could be exported via an OAI-PMHinterface,whichprovidesRDF/XMLusingCIDOCCRM.TheotherdataexportsweretransformedtoXMLandimportedintoFORTH-ICS’MappingMemoryManager.The3MeditorwasusedtodescribethedatasetswithCIDOCCRMandtransformthedataintoRDFformat.ToenableaunifiedsearchenvironmentforalldatasetsitwasalsonecessarytoharmonizedifferingCIDOCCRMmappingsaswellasmaptermstoacommonreferencevocabulary,e.g.archaeologicalterminologytotheAATandplacestotheiDAI.gazetteer.The Linked Data has been stored in a Blazegraph graph database (triple store) to performarchaeologicallyrelevantSPARQLqueriesonthedatatoshowcasethepossibilitiesoftheapproach.Thesearch interfacehasbeen implementedwithMetaphactsontopof theBlazegraphtriplestoreandallowsaccessingthedatainawikisystem.An object-centric and a sites-based view into the cloud of archaeological linked data have beenexplored. The research questions in the object-centric view concerned comparable objects byapplyingthesameparameters.Forexampleoneobject-centricquerywasaboutafragmentaryheadofaSatyrthatwasfoundinChimtou.Thesites-basedviewconcernedquarries,forexamplequarrieswherewhitemarblewasproduced.Heresearchquestionswereaboutallpossiblesculpturesfromaspecificquarry(Pentelli),andliteraturethatdescribesobjectswhicharemadeoutofthemarbleofthatquarry.TheapproachdemonstratedtheadvantagesoftheextendedCIDOCCRMforresearchasqueriestoanswerarchaeologicalquestionscouldberunsuccessfullyovertointegrateddatasets.

235FRBRoomodel,v2.1,February2015,http://www.cidoc-crm.org/frbr_drafts.html236BasicGeo(WGS84lat/long)Vocabulary,https://www.w3.org/2003/01/geo/237Arachne,thecentralobjectdatabaseoftheGermanArchaeologicalInstituteandtheArchaeological

InstituteoftheUniversityofCologne,http://arachne.uni-koeln.de238DeutschesArchäologischesInstitut,Simitthus/Chimtou(Tunesien)Projekt,

http://www.dainst.org/projekt/-/project-display/33904239BritishMuseum-SemanticWebCollectionOnline,http://collection.britishmuseum.org240OxfordRomanEconomyProject(OxfordUniversity):StoneQuarriesDatabase

http://oxrep.classics.ox.ac.uk/databases/stone_quarries_database/241AgoraExcavations(AmericanSchoolofClassicalStudiesinAthens),http://agora.ascsa.net



Thewoodenmaterialdemonstrator

Thewoodenmaterial demonstrator is beingdevelopedbyUniversity of SouthWales (HypermediaResearchGroup) in collaborationwithADS,DANSandSND. It aims to investigate thepotential forNatural Language Processing information extraction techniques to achieve a degree of semanticinteroperabilitybetweenarchaeologicaldatasetsandthetextualcontentofgrey literaturereports.Thus the aim is to extract more specific information from the reports than is available in themetadata alone. Similar NLPmethods will be employed to those used in the Coins demonstratordescribedabove.TheworkbuildsonthetechniquesdevelopedfortheUKSTARProject(Tudhopeetal.2011b;Vlachidisetal.2015).OutputwillbeexpressedasRDFusingthesameCIDOCCRMmodelasusedfortheCoinsDemonstratorwithmappingsmadetotheAAT.Thecasestudyhasabroadthemerelatingtowoodenmaterialincludingshipwrecks,withafocusonindications of types of wooden material, samples taken, wooden objects with dating fromdendrochronological analysis, etc. The work is ongoing and will be reported in the forthcomingARIADNEdeliverableD15.3 (ARIADNE 2017b). The intention is to drawon both English andDutchlanguagedatasetsandgreyliteraturereports,togetherwithSwedisharchaeologicalreports.TheendresultwillbeaSPARQLpilotdemonstratorofthetechnicalpossibilities,operatingoveraLinkedDataexpressionoftheoutput,whichwilloffercrosssearchoverboththedatasetsandtextreports.Itisintended that the demonstrator will explore possibilities for a more (archaeology) user-centredapplication interface (using the ‘widget’ techniques developed in the SENESCHAL project) than aplainSPARQLendpoint.

7.7 Briefsummaryandlessonslearned

Briefsummary

The developmental ARIADNE Linked Data work described in this chapter has focused on theproduction of (and support for) SKOS subject vocabularies,mappings between those vocabulariesandtheArt&ArchitectureThesaurus,inordertoprovideamultilingualcapability,andthemappingsof datasets to the CIDOC-CRM. Furthermore three advanced case studieswith demonstrators arepresentedthatgenerateanduseLinkedDatabasedontheCIDOCCRMandkeysubjectvocabularyhubs:coins,woodenmaterialandsculptures.

The first two case studies involve information extraction from text reports in addition tomappingdatasets, while the third explores external linking beyond the immediate ARIADNE datasets.ExploratoryworkonminingofLinkedDataandNLPtechniquesaredescribedbutbothareresearchareas with potential for much further work. The transformation of the metadata of the datasetsregisteredintheARIADNEdatacataloguetoLinkedDataisdescribedinthenextchapter,asarethedetailsoftheARIADNELinkedDataservice.

The demonstrators are still being finalised at the time of this deliverable butwill be available forgeneraluseviatheARIADNEPortal.Forthereasonsdiscussedintheearlychapters,thecasestudiesareexperimentalinvestigationsofthefutureusecasesthatareaffordedbyLinkedDatatechnology;they result in (working) research demonstrators rather than actual operational systems. Theyillustrate thekindsofpossibilities forcrosssearchandthesemantic integrationofdiversekindsofdatasetsandtextreportsthatLinkedDataandtherelatedsemantictechnologiesmakepossible.

Oneobviousfindingfromtheexperiencetodateisthecriticalimportanceofthesubjectvocabularies(e.g.theAAT)combinedwiththeCIDOCCRMontologyentities,whichactaslinkinghubsinthewebofdata.Moreworkisneededontheidentificationoffurtherlinkinghubsandconsequentsemantic



enrichmentoftheLinkedDatatorelevantexternaldatasets.Oneexampleofapotentiallinkinghubis thePeriod0setofculturalperiodswhichcanbeusedbyprovidersofvariousarchaeologicalandotherculturalheritagedatasets.

Necessary for thewidespreaduptakeof theLinkedDataapproach is theavailabilityofavarietyofmapping and alignment software for different contexts, together with evaluative studies andguidelinesastotheiruse.Beyondthat,tomotivateuserorganisationstodevotescarceresourcestoworkingwithLinkedData,someexemplarworkingapplicationsareneededthataddressarealuser(scientific/research)need.Suchapplicationsshouldofferauserinterfacethatiseasyandattractivetoworkwith,onethatdoesnotrequireprogrammingskillsordetailedknowledgeoftheunderlyingdataschemaorontologystructure.

Itshouldnotnecessarilybeassumedthattheend-applicationdirectlyoperatesovera(LinkedData)triplestore.Thereareadvantagesindoingsofordataupdatesandexternalconnectionsanditisanobviousroute.However,periodicharvestingofLinkedDataisapossibilityforapplicationsthathavereasons toemployawider rangeofprogrammingplatforms.Anotherpossibility is for LinkedDataproviderstoconsiderexposingprogrammaticwebservicesforapplicationdevelopers(inadditiontoa SPARQL endpoint), assuming that an appropriate set of of use cases for the services can beidentified.

Lessonslearned

o MappingofdatasetstoestablisheddomainKOSs(inourcaseCIDOCCRM,AATandothers)allowstheirintegrationwithinandbeyondthecatalogueofadataportal.

o State-of-the-art linking hubs will play an increasingly important role in the web of LOD,comprehensivedomainthesauriastheAATaswellasspecialisedvocabulariesliketheNomismathesaurus.

o The mapping of datasets to such hubs requires domain knowledge, easy to use tools, andguidance of users who carry out such work for the first time. While recommender tools arehelpful,fullyautomatedmappingappearsunlikelytoachivequalityresultsatthecurrenttime.

o TheARIADNEportalandpilotdemonstratorsshowthatthisworkisworththeeffort.Butthereisstill away to go before advanced uses of LODwill become applicable and beneficial in onlineresearchenvironments;moreeffortmustbeinvestedtomakethishappen.

o ThereismuchscopetoexploretheutilityofLODinpractice,takingaccountoftheobjectivesandrequirementsofdifferentusercommunities.ThebestwaystoprovideandemployLODwilllargelydependon their specific contexts (museumcollections, dataarchivesor researchplatforms, forinstance), together with the anticipated use cases. In order to motivate user organisations towork with Linked Data, exemplar working applications that address a real user(scientific/research)needwouldbeveryhelpful.



8 ARIADNELODCloud

8.1 TheARIADNELODCloud–inbrief

TheARIADNELinkedOpenDataCloud(ALDC)isawebofdatathatencompassesrelevantvocabularypartsofthewiderLODcloud,suchastheCIDOCCRM,Art&ArchitectureThesaurus(AAT),nationaland other vocabularies as well as instance data of archaeological and other cultural heritagedatasets. The core linking “hubs” are the CIDOC CRM and AAT as they are themain vehicles forlinkingto/fromtheARIADNEcataloguemetadata.

The ARIADNEmetadata repository is an integrated semantic network, an aggregation of the dataproduced through the process of mapping and transformation of each data provider’s sourcedatabase to the common target ARIADNE CatalogueDataModel (ACDM). Furthermore the ACDMhasbeenmappedtotheCIDOCCRMtoenableapplicationsthatemploycatalogueinformationanditemlevelinformationofvariousdatasets,forexamplesetsofLinkedDatawithCIDOCCRMmappingof the pilot demonstrators. The various Linked Data generated in the project, including links toexternalresources,isbroughttogetherinaLinkedDatagraphdatabasewhichformsthebasisoftheARIADNE LODCloud (ALDC). Thedatabase content is accessible via a SPARQLendpoint to internalandexternalapplicationdevelopers.

ThereareseveralreasonsforbringingtogetheralltheavailabledataintheALDC:

o Shareability:ByusingdefactostandardssuchasthosepromotedbytheW3CundertheumbrellaoftheSemanticWeb,thedataintheARIADNEinformationspacearemadeuniversallyaccessiblefromauniquepoint.

o Interoperability: ByusingCIDOCCRM thedata in theARIADNE information spacearemadeasinteroperableaspossible.CoupledwiththetechnicalinteroperabilitysupportedbytheSemanticWeblanguages(RDF,RDFS,SKOS),thissemanticinteroperabilityprovidesmaximumre-usability.

o Scientificdiscovery:Besidesthetworeasonsabove,theALDCrepresentsanattemptofbringingtogether several kinds of archaeological data, related by subject, temporal and geo-spatialoverlapping. These datapotentially enable scientists to address research questions that couldnot be addressed based on the individual resources. As will be discussed in due course, thispotentialisbeingexploredtoseewhetheritcanactuallyprovidenewscientificknowledge.

ItmustbestressedthatthecurrentALDCistheinitialstageofaninformationspacethatisexpectedtogrowintermsofdata,vocabularies,servicesandusers.TheroleoftheARIADNEprojecthasbeentosetupthisinformationspaceandtoendowitwithafirstportfolioofvaluabledata,vocabulariesandservices.But, if really successful, theALDCwillneverbecompleted.Rather, itwill continuetogrowandevolve,reflectingthegrowthandtheevolutionofLinkedDatagenerationandusagebythearchaeologicalresearchanddatamanagementcommunity.

Thenextsectionsareorganisedasfollows:FirsttheALDCarchitectureisintroduced,highlightingthelogical components that make up the overall system. Each component is then described in thesubsequentsections,emphasizingthecontentofthecomponentintermsofdata,vocabulariesandmappings. Furthermore the strategy followed to make the ALDC discoverable on the web ispresented. The final section summarises and provides some lessons learned in the work on theARIADNELODCloud.



8.2 Architecture

Figure1presentsthearchitectureofARIADNELODCloud(ALDC)inasimplified,diagrammaticform:

Figure1:ArchitectureoftheARIADNELODCloudsystem

Thearchitectureisshownwithinthelargestboxlabelled“ARIADNECloud”.ItcomprisesofhardwareandsoftwarecomponentsthattogetherrealizetheALDC.TheservicesoftheALDCcanbeaccessedintwodifferentways,indicatedintheFigurebytheboxesoutsidethe“ARIADNECloud”:

o HumanscanusetheLinkedDataSectionof theARIADNEPortal,whichenablesthemtoobtainvocabularies andmappings,use theCIDOCCRMbased LinkedDatademonstrators, andaccessdataviaaSPARQLinterface;

o Software agents can use the Linked Data API to issue SPARQL queries against the underlyingtriplestore,therebyobtainingtherequesteddatainoneoftheformatssupported.

ThearchitectureoftheALDCconsistsofthefollowingcomponents:

o D4SciencePlatform: TheD4SciencePlatform isahybriddata infrastructureoffering services tosupport the activity of researchers. At present it connects 2500+ researchers in 44 countries,integrating over 50 heterogeneous data providers. With 99.7% service availability it providesaccess to over a billion records in repositoriesworldwide and executes over 13,000models&algorithms permonth. In the context of ARIADNE, the platform is being used for running thesemantic technologies that support theALDC (triple storeandSPARQLEngine). It also relievestheALDCdevelopersfromtheburdenofimplementinglow-levelservicessuchasauthentication,memorymanagement, security and the like. In addition, the platform allows easy installation,configuration,managementandoperationof theDemonstrators. Finally, itoffersadistributedandscalable file system,accessible throughauser-friendly interface, forhostingandaccessingdatathatarenotingestedinthetriplestores,suchasmappings.

o SPARQL engine and RDF triple store: The semantic technologies employed by the ALDC are aSPARQLengineandanRDFtriplestoreoperatedbytheSPARQLengine.ThesearedeployedonavirtualmachineinstalledonandoperatedbytheD4Scienceplatform.ThetriplestorehoststhedatasetsincludedintheALDC,alongwiththeontologiesdefiningtheclassesandpropertiesused

D4SciencePlatform

RDF TripleStore

SPARQLengine

ARIADNECloud

Mapping&OntologyServer

Demonstrators

L.O.Data Server

LinkedData API

LinkedData Section

ARIADNE Portal



inthesedatasets.ThetechnologyemployedforthesetwocomponentsistheVirtuosoUniversalServer,initsopen-sourceedition242andtheBlazegraphgraphdatabase243.

o TheservicesfortheusersoftheALDC,whetherhumansorsoftwareagents,areofferedbythefollowingcomponents:

- LinkedOpenData Server: Provides access to theARIADNE LinkedDatawhich comprisesofARIADNEcataloguedata(basedontheACDM,whichisalsomappedtotheCIDOCCRM)anddataoftheDemonstrators (seebelow).Theserver is technically implementedasaSPARQLendpoint,endowedwithaprogrammaticandanend-userinterface.BothinterfacesreceiveSPARQLqueries,executethosequeriesagainsttheunderlyingSPARQLEngine,andreturntheresultstotheuserintheappropriateformat,dependingontheselectedaccesschannel.

- Demonstrators:ExemplifythecapabilityofLinkedDatabaseditem-leveldataintegrationtosupportansweringarchaeologicalresearchquestions.Theyrepresentthreedifferentsubjectareasofarchaeology:coins,sculpturesandwoodenmaterial.ForeachanumberofdatasetshavebeenintegratedbasedonmappingstotheCIDOCCRM(andrecentextensions)anduseofotherdomainvocabularies.

- MappingandOntologyServer: Isa file system-like interface forbrowsinganddownloadingthemappingsandtheontologiesinvolvedintheALDC.Thisinterfaceisexclusivelyforhumanusers and accessible from a Virtual Research Environment implemented on top of theD4Science platform. The interface is being provided for the sole purpose of browsing andaccessing mappings and ontologies, while the service for discovering such resources isofferedbytheLinkedOpenDataServer.

Adetaileddescriptionofthecontentsofeachcomponentisgivenbelow.

Fromatechnicalpointofview,theALDCarchitectureincludesmanyothercomponents,requiredfortheproperoperationsofthose listedabove.TheD4Scienceplatformitself includesdozensofopensourcecomponents,whichareintegratedintotheplatform.Butthesecomponentsarenotshownastheyimplementinternalservicesnotdirectlyperceivedbytheusersandassuchoutsideofthescopeofthispresentation.

8.3 TheLinkedOpenDataServer

TheARIADNELinkedOpenDataServer runsa largeRDFdataset, consistingof severalRDFgraphs,eachcorrespondingtoanarchaeologicaldataset.Allgraphsareexpressed inthevocabularyof theCIDOCCRM,includingrecentextensionsoftheontology.Themaindatasets(graphs)arethedatasetoftheARIADNECataloguerecordsandthedatasetsoftheDemonstrators.

ARIADNECataloguedataset

o This dataset contains the data of all catalogue records, expressed in RDF and based on twodifferent vocabularies: the ARIADNE Catalogue DataModel (ACDM) and the CIDOC CRM. TheACDM-basedrecordsdescribethedataresourcesthatarebeingmadeaccessiblebytheARIADNEdataprovidersthroughtheARIADNEPortal.ThesedescriptionshavebeendirectlyimportedfromtheMORedataaggregationinfrastructuresupportingtheARIADNECatalogueservice.TheCRM-based versions of the descriptions have been generated by first creating the ACDM to CRM

242https://virtuoso.openlinksw.com243https://www.blazegraph.com/product/



mappingsand thenapplying thosemappings to theACDM-baseddescriptions. TheCRM-baseddescriptionshavebeenproducedtoenableahigherdatainteroperability,asisdemonstratedbyoneofthedemonstratorsintheALDC(seetheCoinsdemonstratorbelow).

o InadditiontotheACDM/CRM-baseddescriptionsofthecataloguerecordstherearedescriptionsof datasets resulting from the item-level integration of datasets generated and used by theDemonstrators;thesedescriptionsarealsoexpressedinACDM-CRM.

ARIADNEDemonstratorsdatasets

In addition to the catalogue-level data, the LinkedOpen Data Server includes the datasets of theDemonstrators. Here we feature only the datsets of the three main Demonstrators (Coins,Sculptures,WoodenMaterial),whicharebrieflydescribedinthenextsection.Descriptionsofotherdemonstrators, and the datasets used by them, are given in the D14.2 Pilot DeploymentExperiments.

o Coins demonstrator: This dataset results from the item-level integration of information aboutcoins from five datasets based on the CRM, Nomisma ontology, and Art & ArchitectureThesaurus. The demonstrator employs the core CRM, the extension CRMdig and a small coin-specificextensionmodellingcategoricalinformation.Theintegrateddatasetsare:

- dFMRÖ - Digitale Fundmünzen der Römischen Zeit in Österreich (Digital Coin-finds of theRomanPeriodinAustria),isarelationaldatabaseofpre-RomanandRomanImperialperiodcoins found in Austria and Romania (75,565 records of coin finds), developed by theNumismaticsResearchGroupattheAustrianAcademyofSciences;

- MuseiD-Italia documentation of several coins collections of Italianmuseums integrated inCulturaItalia;

- Asubsetofnumismaticsrecords(1670)fromtheFitzwilliamMuseum(Cambridge)databasefromtheCOINSproject(2007-2009,ledbyPIN);

- Coinsdatarecords(630)fromtheSoprintendenzaArcheologicadiRoma(SAR)database,alsofromtheCOINSproject;

- Documentation of coin finds (517) in the iDAI.field research database of the Pergamonproject,withdetailedinformationaboutthearchaeologicalcontext;

- The result of knowledge extraction using Natural Language Processing methods from acollectionoftextualdocumentsaboutcoins.

o Sculpturesdemonstrator:AsetofdatafromfivedifferentdatabasesbasedontheCRM,CRMsciandCRMarchaeo,usingtheBasicGeovocabularyandtheobject-orientedversionofFunctionalRequirements for Bibliographic Records (FRBRoo) for describing bibliographical records. Thedatasetcomprisesofsculpturesdatafrom:

- BritishMuseum:SemanticWebCollectionOnline(ismappedtothecoreCRMandincludeslinkstoBMvocabularies),wasaccesseddirectlyviaitsSPARQLendpointsandintegratedbyusingaSPARQLfederatedquery;

- Arachne, data exported via an OAI-PMH interface, which provides RDF/XML using CIDOC-CRM;

- iDAI.fielddatabaseoftheChimtouproject,transformedtoXMLandimportedintoFORTH’s3Mtool,describedwithCIDOC-CRMandtransformedtoRDF;



- OxfordRomanEconomyProject:StoneQuarriesDatabase,RDFgenerationasabove;

- Athenia Agora excavation DB (over 280,000 data items), mapped using the extensionsCRMarchaeo and CRMsci; the most relevant parts of the database schema have beenmappedtoCRM,alsousingCRMarchaeoandCRMsci.

o Wooden Material demonstrator: A dataset with a broad theme relating to wooden materialincluding shipwrecks,with a focus on indications of types ofwoodenmaterial, samples taken,woodenobjectswithdatingfromdendrochronologicalanalysis,etc.Thedatahasbeenextractedfrom archaeological datasets and grey literature reports in different languages and expressedusingtheCIDOCCRMandmappingsmadetotheAAT.Theintegrateddatasetsare:

- Digital Collaboratory for Cultural Dendrochronology (DCCD) dataset, an extract of theinternationalDCCDdatabasefacilitatedbyDANS;

- DendrochronologyDatabaseof theVernacularArchitectureGroup (UK),2016.ArchaeologyDataService(doi:10.5284/1039454);

- Cruck Database of the Vernacular Architecture Group (UK), 2015. ADS (doi:10.5284/1031497);

- Newport Medieval Ship. N. Nayling (Univ. Wales Trinity St David) & T. Jones (NewportMuseumsandHeritageService),2014.ADS(doi:10.5284/1020898);

- Mystery Wreck Project (Flower of Ugie). Hampshire and Wight Trust for MaritimeArchaeology,2012.ADS(doi:10.5284/1011899);

- DataextractedviaNLP from25archaeologicalgrey literaturereports inDutch,EnglishandSwedish(reportsprovidedbyADS,DANSandSND).

The rationale for uniting all datasets, the datasets of the ARIADNE Catalogue, the mainDemonstratorsandothersintheARIADNELODCloudistwofold:theaccessibilityoftheLODdatasetsfromasinglesourceisclearlyanadvantageforresearchers,andthereistheambitionofsupportingresearchquestions inarchaeologythatcouldnotbeaddressedbasedon individualcollections.TheDemonstratorsarefirstexperimentsonthediscoveryofknowledgeacrossseveraldifferentdatasets;theexperimentationisongoing.

Connections

ThereexistseveralconnectionsamongsttheLinkedDatagraphsaddressedabove.AllCatalogue-leveldata are expressed in the same vocabularies (ACDM, CIDOC CRM), and link to the same externalLinkedDatavocabularies.This includestheSKOSversionoftheArt&ArchitectureThesaurus(AAT)which isemployedas thebackboneof theARIADNEsubjects terminology“hub”.Other thesauri inSKOSformatareinvolvedthroughthemappingoftermsusedindataproviderrecordstotheAAT,forexample, themulti-lingual PACTOLS thesaurus andHistoric England thesauri. Figure 2 presents anACDMbasedCatalogue-leveldescriptionofacoindatasetusingAATconcepts.

<rdf:RDFxmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">

<rdf:Descriptionrdf:about="http://schemas.cloud.dcu.gr/#acdm:ariadne/acdm:ariadneArchaeologicalResource/acdm:dataset">

...

<rdf:Descriptionrdf:about="http://.../acdm:dataset/acdm:ariadneSubject">



<rdf:Descriptionrdf:about="http://.../acdm:dataset/acdm:ariadneSubject/acdm:derivedSubject">

<skos:prefLabel>coins(money)</skos:prefLabel>

<dc:source>http://vocab.getty.edu/aat/300037222</dc:source>

</rdf:Description>

</rdf:Description>

<rdf:Descriptionrdf:about="http://.../acdm:dataset/acdm:ariadneSubject_2">

<rdf:Descriptionrdf:about="http://.../acdm:dataset/acdm:ariadneSubject_2/acdm:derivedSubject">

<skos:prefLabel>archaeologicalsites</skos:prefLabel>

<dc:source>http://vocab.getty.edu/aat/300000810</dc:source>

</rdf:Description>

</rdf:Description>

Figure2:ExampleofanACDM-baseddescriptionofadataset

All item-level data of the demonstrators are expressed in the CIDOC CRM vocabulary, and link toexternal vocabularies employed by the demonstrators. For example, terms in coins datasets arelinkedtotheNomismathesaurusortoponymsinsculpturesdatasetsarelinkedtotheiDAI.gazetteer.Demonstratorsalsouseexternaldatasets,forexamplethesculpturesdemonstrator linkstodataintheBritishMuseum’sSemanticWebCollectionOnline.

Catalogue-levelanditem-leveldataarelinkedtoeachotherbyemployingspecificpropertiesoftheCIDOCCRM.Forexample,coindataarelinkedtoARIADNEcataloguerecordsbyaddingtoeachcoina triple linking it to the datasetwhere the information about the coin belongs. This connection isestablished through the CRM property P67i_is_referred_to_by. The type of the triple thatimplementsthelinkingbetweenacoinrecordandanACDMrecordis:

Thecoin(subject): E22_Man-Made_Object->

TheCRMproperty(predicate) P67i_is_referred_to_by->

TheACDMrecord(object): E73_Information_Object

Moreover, NLP results are linked to the coins through terms of the Nomisma.org vocabulary andthen to theARIADNE catalogue records through the linksbetween coins and recordsas describedabove.

In thisway information in the cataloguedataset is integratedwithotherdatasets (e.g.datasetsofcoins, wooden material, sculptures, etc.) allowing to query the Linked Data at different levels ofinformation,catalogueinformationaswellasitemspecificinformation.

TogivesomefiguresofthecurrentARIADNELODCloud:ThedatasetoftheARIADNEcataloguehas20+millionRDF triples, theCoinsdemonstrator 1+million triples, the Sculpturesdemonstrator 5+milliontriples,andtheWoodenMaterialdemonstrator1+milliontriples.Theingestedvocabulariesamount to4+million triplesofwhich theAAT is the largestpart. Thus theARIADNELODCloudatpresentcontainsatotalofabout32milliontriples.



8.4 TheDemonstrators

The Demonstrators represent three different subject areas of archaeology, coins, sculptures andwoodenmaterial.ThedatasetsthatarebeingemployedbytheDemonstratorsaredescribedabove.Thedatasetshavebeenharmonized,wherenecessary,usingtheCIDOCCRM(andrecentextensions),transformed into RDF graphs and ingested into the ARIADNE LOD Cloud. The Demonstrators aredescribed in greater detail in the deliverable D14.2 Pilot Deployment Experiments and thedeliverableD15.3SemanticAnnotationandLinking.

TheDemonstratorswillbecomeaccessibletoend-usersthroughadedicatedLinkedDataSectiononthe ARIADNE Portal. They have been developed to exemplify the capability of Linked Data baseditem-level data integration to support answering archaeological researchquestions. This capabilitybuildsonthemappingofdatasetstotheCIDOCCRM(includingrecentextensions)andotherdomainvocabularies(i.e.AAT,Nomismaandothers).Herewegiveabriefaccountofsomepromisingresultsthathavebeenobtainedindemonstrators.

TheCoinsDemonstratorcanillustrateimportantpointsthatarepresentalsoinotherdemonstrators.The Coins Demonstrator employs datasets of different providers (including results of NLP ofarchaeological grey literature), mappings to the CIDOC CRM (and CRMdig extension), and otherdomain vocabularies (AAT, Nomisma). Furthermore it presents a case that shows the potential ofquerying,intheARIADNELODCloud,thisitem-leveldatatogetherwithcatalogue-leveldata.

QueriesacrossthedatasetsoftheCoinsDemonstratorshowusefulresultsforresearchers.Queriesthatare trivial tobeansweredbyeachdataset separatelybecomerelevant fora researcherwhentheyareexecutedacrossseveraldatasets,andtheresultscombinedbytheresearcher.ForexamplesearchessuchasFindcoinsmintedinthesameplace/area,Findcoinsmintedbythesameauthority(e.g.Antonianus),Findcoinsproduced in thesameperiod (e.g. thesamecentury),Findcoinsmadefrom specific material (e.g. bronze), etc. Moreover, item-level and catalogue-level data can bequeriedsimultaneously,e.g.Findthepublishersofallcollectionsthatcontainbronzeantoninianus.

The Sculptures Demonstrator has the same general characteristic but involves some differentaspects. For example, the datasets include data from excavations and instead of grey literaturereports the large Zenonbibliographic databaseof theGermanArchaeological Institute is involved.Consequently the SculpturesDemonstrator employs the CRMextensions CRMarchaeo andCRMsciandFunctionalRequirementsforBibliographicRecords(FRBRoo),alongwithothervocabularies(e.g.the AAT and the iDAI.gazetteer). Also this demonstrator shows advanced capability to supportanswering archaeological research questions. For example, queries over the datasets concernedquarries where white marble was produced, all possible sculptures from a specific quarry, andliteraturethatdescribesobjectswhicharemadeoutofthemarbleofthatquarry.

ThewoodenmaterialDemonstratoralsosharesthegeneralcharacteristicswithaparticularfocusonthe integration of grey literature textual reports in different languages with datasets on adendrochronological theme. The complexity of the underlying semantic framework based on theCIDOC CRM and Getty AAT is shielded from the user by theWeb application user interface. TheDemonstratorhighlightsthepotentialforarchaeologicalresearchthatcaninterrogategreyliteraturereports inconjunctionwithdatasets.Queriesconcernwoodenobjects(e.g.samplesofbeechwoodkeels),optionallyfromagivendaterange,withautomaticexpansionoverhierarchiesofwoodtypes.



8.5 TheMappingandOntologyServer

TheMapping andOntology Server provides information about themappings and the vocabularies(ontologies,thesauri)involvedintheARIADNELODCloud.

ThefollowingmappingsofdatasetstotheCIDOCCRM(andextensions)areavailable:

o SchemasoftheItalianCentralInstituteforCatalogueandDocumentationforarchaeologicalfinds(RA)andmonumentsandcomplexes(MA/CA)mappedtotheCRM,using,whererequired,morespecialisedclassesandpropertiesofCRMextensions(providedbyICCU);

o Database schema and concepts of SITAR, the Archaeological Territorial Informative System ofRomemappedtotheCRMandCRMarchaeo(ICCUincooperationwithotherinstitutions);

o dFMRÖ(coinsdatabase)mappedtoCRM,CRMdigandaspecializedextensionforcoins,usedintheCoinsdemonstrator(ÖAW);

o iDAI.fielddatabaseofthePergamonprojectmappedtoCRM,CRMarchaeoandCRMsci,usedintheCoinsdemonstrator(DAI);

o iDAI.fielddatabaseof theChimtouproject includingstoneobjectsandarchaeological contexts,mappedasaboveandusedintheSculpturedemonstrator(DAI);

o AtheniaAgoraexcavationdatabase(over280,000dataitems),mappedasaboveandusedintheSculpturesdemonstrator(DAI);

o Digital Collaboratory for Cultural Dendrochronology (DCCD) dataset, an extract facilitated byDANS,mappedtotheCRM(USW);

o Dendrochronology Database of the Vernacular Architecture Group (UK), 2016 (doi:10.5284/1039454),providedbyADS,mappedtotheCRM(USW);

o Cruck Database of the Vernacular Architecture Group (UK), 2015 (doi: 10.5284/1031497),providedbyADS,mappedtotheCRM(USW);

o NewportMedievalShip.N.Nayling&T.Jones,2014(doi:10.5284/1020898),datasetprovidedbyADS,mappedtotheCRM(USW);

o MysteryWreckProject(FlowerofUgie).HampshireandWightTrustforMaritimeArchaeology,2012(doi:10.5284/1011899),datasetprovidedbyADS,mappedtotheCRM(USW);

o AnimalBoneEvidenceSouthEngland(doi:10.5284/1000102),datasetprovidedbyADS,mappedtotheCRMandextensionsandusedinanAnimalRemainsdemonstrator(DAI);

o Holozängeschichte der Tierwelt Europas (doi:10.13149/001.mcus7z-2), dataset provided byIANUS,mappedandusedasabove(DAI).

Thefollowingontologiesareavailableasreferences:

o CIDOCCRMcore.Version5.0.4,December2011;

o CRMarchaeo. Model for integrating metadata about the archaeological excavation process;introducesconceptsofstratigraphyandexcavation.Version1.4,April2016;

o CRMsci. Model for integrating metadata about scientific observation, measurements andprocesseddata.Version1.2.3,April2016;



o CRMdig.Modelofdigitisationprocesses, toencodemetadataabout the stepsandmethodsofproduction(“provenance”)ofdigitalrepresentationssuchas2D,3Doranimatedmodels.Version3.2.1,April2016;

o CRMba.Modelforinvestigatinghistoricandprehistoricbuildings,therelationsbetweenbuildingcomponents,functionalspaces,topologicalrelationsandconstructionphasesthroughtimeandspace;harmonizedwithCRMarchaeo.Version1.4,April2016;

o CRMgeo.Spatio-temporalmodelthatintegratesCRMandOGCstandards.Version1.2,February2015;

o CRMinf. Model for integrating data with scholarly argumentation and inference making indescriptiveandempiricalsciences;harmonizedwithCRMsci.Versionv0.7,February2015;

o FunctionalRequirementsforBibliographicRecords,FRBRooencoded inRDFS.Version2.4, June2016.

ThefollowingthesauriinSKOSareavailableasreferences:

o AAT-Art&ArchitectureThesaurus(Getty);

o PACTOLSthesaurus(Peuples, Anthroponymes, Chronologie, Toponymes, Œuvres, Lieux etSujets) of the Fédération et ressources sur l’Antiquité, France. A largemulti-lingual thesauruswhich focuses on antiquity and archaeology from prehistory to the industrial age; terms inFrench, English, German, Italian, Spanish, Dutch, and (some) Arabic). Over 1600 PACTOLSconcepts,usedbyInrapintheircatalogueofarchaeologicalreports(DOLIA),havebeenmappedtotheAAT;

o HistoricEnglandthesauri(ForumonInformationStandardsinHeritage–FISH),thesauriinSKOSprovidedbyHeritageData(SENESCHALproject).ADS,employsfiveofthethesauri(monuments,components, building-material,maritime-craft, fishobjects) ofwhichabout850 conceptshavebeenmappedtotheAAT;

o PICOthesaurus(ICCU):Alargethesaurusoftermsrelatedtocultureandculturalheritage(Italianand English) which is being used for the data of CulturaItalia; a number of terms concernarchaeologywhichhavebeenmappedtotheAAT;

o Italian Archaeological Finds Vocabulary / Reperti Archeologici (RA) Thesaurus, a thesaurusdescribingarchaeologicalfinds(ICCU);

o RCE Archeologisch Basisregister - ABRr+ thesauri (Rijksdienst Cultureel Erfgoed, Netherlands),about450conceptsofmonument types (Archeologischecomplextypen)havebeenmappedbyDANStotheAAT;

o IrishMonument Types thesaurus (NationalMonuments Service), a hierarchical list of conceptsexpressedinSKOSaspartoftheLoCloudproject;

o iDAI.vocab: group of 14 thesauri of archaeological terminology in different languages and ofvariedsize;theGermanthesaurus,mappedtotheAAT,servesasthecentralhubtoandthroughwhichtheotherthesauriarelinked;

o iDAI.Gazetteer:providesover1millionentriesdescribingmodernandancientplacesthatareofinterest tothearchaeologistsandalsoactsasahubby linkingothergazetteers likeGeonamesandPleiades;

o Dendrochronology multi-lingual vocabulary of the Digital Collaboratory for CulturalDendrochronology,developedandrecentlyexpressedinSKOSbyDANS;



o EAGLEepigraphy vocabularies (Material, Typeof inscription, Execution technique,Object type,Decoration,Datingcriteria,Stateofpreservation);

o Nomismaontologyofnumismaticconceptsandentities(Nomisma.org).

8.6 Promotionofexternaluse

One of the core principles of Linked Open Data is linking of published datasets to others whichgenerates an expanding and increasingly rich web of Linked Data. Promotion of linking relevantdatasets to the ARIADNE LOD by external developers is planned to include documentation of thedatainrelevantregistries,targeteddisseminationofinformationabouttheavailabledata,anddirectdiscussionwithanumberofinteresteddevelopers.

Data registration: Documenting sets of LOD in relevant registries makes it easier for applicationdeveloperstoidentify,evaluateandlinktorelevantdatasets.TheVocabularyofInterlinkedDataSets(VoID)ismostoftenbeingusedtodescribeandregistersetsofLOD.InVoIDadatasetisacollectionofdata,publishedandmaintainedbyasingleprovider,availableasRDF,andaccessible,forexample,throughaSPARQLendpoint.Figure3illustratesaVoIDdescriptionoftheARIADNELOD:

@prefixrdf:<http://www.w3.org/1999/02/22-rdf-syntax-ns#>.

@prefixrdfs:<http://www.w3.org/2000/01/rdf-schema#>.

@prefixfoaf:<http://xmlns.com/foaf/0.1/>.

@prefixdcterms:<http://purl.org/dc/terms/>.

@prefixvoid:<http://rdfs.org/ns/void#>.

:ARIADNE-LODavoid:Dataset;

dcterms:title"ARIADNEregistry";

dcterms:publisher"ARIADNEProject";

foaf:homepage<http://registry.ariadne-infrastructure.eu>;

dcterms:description"Aregistryofdataforarchaeologicalresearch";

dcterms:license<http://opendatacommons.org/licenses/by/>;

void:sparqlEndpoint<http://ariadne2.isti.cnr.it/sparql>;

…

Figure3:VoIDdescriptionoftheARIADNEregistry

ThefinalARIADNELODwillberegistered intheDataHub(datahub.io),wherealsosomeresourcesemployedbyARIADNEcanbefound(e.g.theGettyAAT,EnglishHeritagethesauri,andothers);otherregistriesandplatforms(e.g.Github,Wikidata)arebeingconsidered.

Targeted dissemination: Announcements and other information about the available LOD will bedisseminatedviarelevantmailinglists,newslettersetc.oftheLinkedDatacommunityinthefieldsofarchaeology,culturalheritage,classicalstudies,historyandotherhumanities.



Directconsultationwithdevelopers:AnumberofLinkedDataapplicationdevelopersof institutionsandprojectswillbecontacteddirectlytosuggestanddiscussinterlinkingwiththeirorotheravailabledatasetsinthewebofLOD.

8.7 Briefsummaryandlessonslearned

Briefsummary

TheARIADNEregistryholdsmetadataofdataresourcesfromthecontentproviders.Thesemetadataare being collected and enriched with an aggregator (MORe) and included in the ARIADNE datacatalogue. ARIADNEmakes the catalogue and other data generated in demonstrators available asLinkedOpen Data (LOD); thereby the ARIADNE LOD can become part of aweb of Linked Data ofarchaeologicalandrelatedotherinformationresources.

ThisworkwithinARIADNEinvolvedtheuseofasuitableRDFstoreandgraphdatabasefortheLinkedData generation and linking efforts. The project has experimented with two such technologies,Virtuoso and Blazegraph, to perform archaeologically relevant SPARQL queries on the generatedLinked Data, and to allow updates of datasets using the SPARQL 1.1 Graph Store HTTP Protocol.Based on this preliminary work, a scalable implementation that can efficiently support thepublication and use of the ARIADNE LOD has been designed and realized to offer three differentservices:theLinkedOpenDataServer,theDemonstrators,andtheMappingandOntologyServer.

The Linked Open Data Server provides access to a large RDF dataset, which comprises of severalgraphs of archaeological datasets and can be queried via a SPARQL endpoint. The DemonstratorshavebeendevelopedtoexemplifythecapabilityofLinkedDatabaseditem-leveldataintegrationtosupportansweringarchaeologicalresearchquestions.Theyrepresentthreedifferentsubjectareasofarchaeology: coins, sculptures and wooden material. For each a number of datasets have beenintegratedbasedonmappingstotheCIDOCCRM(andrecentextensions)anduseofotherdomainvocabularies.TheMappingandOntologyServerprovides informationabout themappingsand thevocabularies(ontologies,thesauri)involvedintheARIADNELODCloud.

ThecurrentARIADNELODCloudis justtheinitialstageofan informationspacethat isexpectedtogrow in terms of data, vocabularies, services and users. Experiments to exploit the ARIADNE LODhave juststarted,withpromisingresultsasshownby theDemonstrators.Planned futureworkwillaim to proceed with linking the available Linked Data to relevant other datasets. To promoteinterlinking, theARIADNE LODwill be announced via relevantmailing lists, newsletters etc. of theLinkedDatacommunity in the fieldofarchaeologyandculturalheritage.AnumberofLinkedDatadevelopers will also be contacted directly to suggest and discuss interlinking with their or otheravailabledatasetsinthewebofLOD.

Lessonslearned

WhiletheLinkedOpenDatastandardsareessentialforintegratingdata,thetechnologysupportingsuchintegrationisstillinitsinfancy.TheARIADNELOD,comprisingofLODoftheARIADNEcatalogue,three demonstrators and various vocabularies sum up to about 32million RDF triples.While anyrelational database can easily handle millions of records, the corresponding amount of RDF in acurrenttriplestorecancauseseriousefficiencyproblemsasexperiencedintheexperimentationwiththe ARIADNE LinkedData Cloud. It is becoming apparent that this is the price to be paid to haveinteroperability. More robust and efficient graph databases are required if we want to proceedtowardsBigDataasLinkedData.ThisisthefirstlessonthatwehavelearnedwhileimplementingtheARIADNELinkedDataCloud.



Thesecondlessoncomesfromthegraphdatamodel.Thismodelisintrinsicallybinary,hencemakesitdifficulttoexpresshigherrankrelations,andtoeasilyimplementdataconnectionpatterns.Inthelatter case, the patterns may involve data chains that span several arcs, and their definition andimplementationisnottrivial.Conversely,correlationsbetweendataitemscanbeepitomizedbysuchpaths,whichneedtobedetected,andthis isacomputationallyvery intensivetask if the lengthofthepathsgobeyond2-3arcs.Thisfacthasalwaysbeenknownfromatheoreticalpointofview,butworkingwithrealdatawecouldexperienceitinpractice.



9 Referencesandrelevantothersources

5★OpenData(detailsBerners-Lee’s5-starschemeofLinkedOpenDatawithexamplesandexplainsbenefitsofandsomeissuesinprovidingsuchdata),http://5stardata.info

Acheson,Phoebe(2014):LinkedOpenBibliographiesinAncientStudies.ISAWPaper7.2,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

AgostiM.,ConlanO.,FerroN.etal.(2013):InteractingwithDigitalCulturalHeritageCollectionsviaAnnotations:TheCULTURAApproach.DocEng’13,Florence,Italy,September10–13,2013,http://www.digitalmeetsculture.net/wp-content/uploads/2013/12/Interacting-with-Digital-Cultural-Heritage-Collections-via-Annotations.pdf

AgriculturalInformationManagementStandards(AIMS):Vocabularies,MetadataSetsandTools(VEST)registry:KOS,http://aims.fao.org/vest-registry/vocabularies

AGROVOCLinkedOpenData,http://aims.fao.org/standards/agrovoc/linked-open-data

AlexanderK.,CyganiakR.,HausenblasM.&ZhaoJ.(2009):DescribingLinkedDatasets.OntheDesignandUsageofvoiD,the“VocabularyofInterlinkedDatasets”.In:ProceedingsoftheLinkedDataontheWeb(LDOW‘09)workshop,Madrid,Spain,20April2009.http://ceur-ws.org/Vol-538/ldow2009_paper20.pdf

AllemangD.&HendlerJ.(2011):SemanticWebfortheWorkingOntologist.EffectiveModelinginRDFSandOWLSecondEdition.MorganKaufmann

AlmasB.,BabeuA.&KrohnA.(2014):LinkedDatainthePerseusDigitalLibrary.ISAWPaper7.3,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

AlmeidaB.,RocheC.&RuteC.(2016):TerminologyandontologydevelopmentinthedomainofIslamicarchaeology,pp.147-156,in:Erdman-ThomsenH.,Pareja-LoraA.&NistrupMadsenB.(2016):TermBasesandLinguisticLinkedOpenData.TKE2016-12thInternationalconferenceonTerminologyandKnowledgeEngineering.CopenhagenBusinessSchool,http://openarchive.cbs.dk/handle/10398/9323

AloiaN.,PapatheodorouC.,GavrilisD.,DeboleF.&MeghiniC.(2014):DescribingResearchData:ACaseStudyforArchaeology,pp.768–775,in:MeersmanR.etal.(eds.):OntheMovetoMeaningfulInternetSystems:OTM2014Conferences.Springer(LNCS8841);preprint,https://www.academia.edu/19889230/Describing_Research_Data_A_Case_Study_for_Archaeology

AmsterdamMuseuminEuropeanaDataModelRDF,http://semanticweb.cs.vu.nl/lod/am

AncientWorldMappingCentre(AWMC/UniversityofNorthCarolina):AntiquityÀ-la-carteandpublicmaptiles,http://awmc.unc.edu/wordpress/alacarte/

AnichiniF.&GattigliaG.(2012):MappaOpenData.Fromwebtosociety.Archaeologicalopendatatesting,pp.54-56,in:OpeningthePast:ArchaeologicalOpenData,MapPapers3-II,http://mappaproject.arch.unipi.it/wp-content/uploads/2011/08/Pre_atti_online3.pdf

AntikeFundmünzeninEuropa(web-basedcoinsdatabasedevelopedbytheRomano-GermanicCommissionoftheGermanArchaeologicalInstitute),http://afe.fundmuenzen.eu



ArbuckleS.,Whitcher-KansaS.,KansaE.,OrtonD.etal.(2014):DataSharingRevealsComplexityintheWestwardSpreadofDomesticAnimalsacrossNeolithicTurkey.In:PLoSONE,9(6):e99845,http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0099845

Archaeogeomancy.net(2014):ColonisationofBritain,30May2014,http://www.archaeogeomancy.net/2014/05/colonisation-of-britain/

ArchaeologyDataService(2015):ADS/InternetArchaeologyAnnualReport,1.8.2014–31.7.2015,http://archaeologydataservice.ac.uk/attach/annualReports/ADS%20Annual%20Report%202014-15.pdf

ArchaeologyDataService:LinkedOpenData,http://data.archaeologydataservice.ac.uk

ArchaeologyDataService:TheSTELLARproject,http://archaeologydataservice.ac.uk/research/stellar/

Archaeotools-Datamining,facettedclassificationandE-archaeology(UK,e-ScienceResearchGrant,2007-2009),http://archaeologydataservice.ac.uk/research/archaeotools

ArcheoInf-InformationszentrumfürdieArchäologie(Germany,DFG-fundedproject,2008-),http://archeoinf.tu-dortmund.de

ArcheologischBasisregister(RijksdienstCultureelErfgoed/CulturalHeritageAgencyoftheNetherlands),http://abr.erfgoedthesaurus.nl

ArcherP.,DekkersM.,GoedertierS.,HarzardN.&LoutasN.(2013):StudyonbusinessmodelsforLinkedOpenGovernmentData(BM4LOGD).StudypreparedfortheISAprogrammebyPwCEUServices,23November2013,https://joinup.ec.europa.eu/community/semic/document/study-business-models-linked-open-government-data-bm4logd

ArchivesHub,http://archiveshub.ac.uk

ArchivesHub:LOCAH-LinkedArchivesandLinkingLivesprojects(2010-2012),http://locah.archiveshub.ac.uk

ARENA-ArchaeologicalRecordsofEurope-NetworkedAccessproject(2001-2004,and2009-2010inthecontextofDARIAH),http://ads.ahds.ac.uk/arena/search/

ARIADNE-LinkedDataSIG(2013):FirstMeeting,EAA2013Conference,Pilsen,4September2013,http://www.ariadne-infrastructure.eu/Community/Special-Interest-Groups/Linked-Data

ARIADNE-LinkedDataSIG(2014):SecondMeeting,CAA2014Conference,Paris,23April2014,http://www.ariadne-infrastructure.eu/Community/Special-Interest-Groups/Linked-Data

ARIADNE(2013):D3.2ReportonProjectStandards(November2013),http://www.ariadne-infrastructure.eu/Resources/D3.2-Report-on-project-standards

ARIADNE(2014a):D2.1FirstReportonUsers’Needs(April2014),http://www.ariadne-infrastructure.eu/Resources/D2.1-First-report-on-users-needs

ARIADNE(2014b):Modelingscientificdata:workshopreport,12September2014,http://www.ariadne-infrastructure.eu/News/Modeling-scientific-data

ARIADNE(2014c):TheWayForwardtoDigitalArchaeologyinEurope.November2014,http://www.ariadne-infrastructure.eu/Media/Files/Ariadne-Booklet-The-Way-Forward-to-Digital-Archaeology-in-Europe

ARIADNE(2015a):D2.2SecondReportonUsers’Needs(February2015),http://www.ariadne-infrastructure.eu/content/view/full/1188



ARIADNE(2015b):D16.1FirstReportonDataMining(March2015),http://www.ariadne-infrastructure.eu/Resources/D16.1-First-Report-on-Data-Mining

ARIADNE(2015c):D16.2FirstReportonNaturalLanguageProcessing(May2015),http://www.ariadne-infrastructure.eu/Resources/D16.2-First-Report-on-Natural-Language-Processing

ARIADNE(2015d):ARIADNEatLinkedPasts:CheckinginonthestateoftheartforLinkedOpenDataandCulturalHeritage.ARIADNEnews,7August2015,http://www.ariadne-infrastructure.eu/News/ARIADNE-at-Linked-Pasts

ARIADNE(2015e):D2.3PreliminaryInnovationAgendaandActionPlan(November2015),http://www.ariadne-infrastructure.eu/Resources/D2.3-Preliminary-Innovation-Agenda-and-Action-Plan

ARIADNE(2016a):D14.1ExtendedCRM(April2016),http://www.ariadne-infrastructure.eu/Resources/D14.1-Extended-CRM

ARIADNE(2016b):D15.1ReportonThesauriandTaxonomies(August2016),http://www.ariadne-infrastructure.eu/Resources

ARIADNE(2017a):D14.2PilotDeploymentExperiments(January2017),willbeavailableathttp://www.ariadne-infrastructure.eu/Resources

ARIADNE(2017b):D15.3ReportonSemanticAnnotationandLinking(January2017),willbeavailableathttp://www.ariadne-infrastructure.eu/Resources

ARIADNECatalogueDataModel(ACDM),http://support.ariadne-infrastructure.eu

ARIADNEDatasetsRegistry,http://registry.ariadne-infrastructure.eu

ARIADNE:AriadneReferenceModel(setofCIDOCCRMextensions,includingreferencedocument,presentation,RDFSencoding),http://www.ariadne-infrastructure.eu/Resources/Ariadne-Reference-Model

AroyoL.,HyvönenE.&vanOssenbruggenJ.(eds.,2007):CulturalHeritageontheSemanticWeb.Proceedingsoftheworkshopco-locatedwiththe6thInternationalSemanticWebConference,Busan,Korea,http://www.cs.vu.nl/~laroyo/CH-SW/ISWC-wp9-proceedings.pdf

ArSol-ArchivesduSol(SoilArchives)project,http://arsol.univ-tours.fr

Arwe,John(2011):CopingwithUn-CoolURIsintheWebofLinkedData.PresentedattheLinkedEnterpriseDataPatternsWorkshop.Data-drivenApplicationsontheWeb,Cambridge,6December2011,http://www.w3.org/2011/09/LinkedData/ledp2011_submission_5.pdf

ASIS&T(2014):SpecialsectionEconomicsofKnowledgeOrganizationSystems.ASIS&T-BulletinoftheAssociationforInformationScienceandTechnology,40(4):13-42,http://asis.org/Bulletin/Apr-14/Bulletin_AprMay14_Final.pdf

AspöckE.&GeserG.(2014):Whatisanarchaeologicalresearchinfrastructureandwhydoweneedit?-AimsandchallengesofARIADNE.In:Proceedingsofthe18thInternationalConferenceonCulturalHeritageandNewTechnologies(CHNT18),Vienna,November2013,http://www.chnt.at/wp-content/uploads/Aspoeck_Geser_2014.pdf

AssafA.&SenartA.(2012):DataQualityPrinciplesintheSemanticWeb.ICSC'12Proceedingsofthe2012IEEESixthInternationalConferenceonSemanticComputing;preprint:arXiv:1305.4054[cs.DL],http://arxiv.org/ftp/arxiv/papers/1305/1305.4054.pdf



AthenaPlus(2013a):FirstreleaseGLAMsectorreferenceterminologies.Projectdeliverable4.1,September2013,http://www.athenaplus.eu/getFile.php?id=187

AthenaPlus(2013b):ReviewonLinkedOpenDataSources.Projectdeliverable4.2,October2013,http://www.athenaplus.eu/getFile.php?id=190

AthenaPlus(EU,CIPBestPracticeNetwork,3/2013-8/2015),http://www.athenaplus.eu

AuerS.,BühmannL.,DirschlC.etal.(2012a):ManagingtheLife-CycleofLinkeddatawiththeLOD2Stack.ISWC2012-11thInternationalSemanticWebConference,Boston,USA,11-15.11.2012,http://iswc2012.semanticweb.org/sites/default/files/76500001.pdf(also:http://svn.aksw.org/lod2/Paper/ISWC2012-InUse_LOD2-Stack/public.pdf)

AuerS.,DemterJ.,MartinM.&Lehmann,J.(2012b):LODStats-AnExtensibleFrameworkforHigh-performanceDatasetAnalytics.ProceedingsoftheEKAW2012–18thInternationalKnowledgeEngineeringandKnowledgeManagementConference,GalwayCity,Ireland,8-12October2012,http://svn.aksw.org/papers/2011/RDFStats/public.pdf

BagosiT.,CalvaneseD.,HardiJ.etal.(2014):TheOntopFrameworkforOntologyBasedDataAccess(OBDA),pp.67-77,in:CSWS2014-TheSemanticWebandWebScience-8thChineseConference,Wuhan,China,8-12August2014,Springer;pre-print,http://www.inf.unibz.it/~calvanese/papers/bago-etal-CSWS-2014.pdf

BarberaN.,MeschiniF.,MorbidoniC.&TomasiF.(2012):Annotatingdigitallibrariesandelectroniceditionsinacollaborativeandsemanticperspective,pp.46-57,in:AgostiM.etal.(eds):DigitalLibrariesandArchives.8thItalianResearchConference(IRCDL2012),CCIS354,Heidelberg.Springer,http://dspace.unitus.it/bitstream/2067/2331/1/paper_annotation_last.pdf

BARTOC-BaselRegisterofThesauri,Ontologies&Classifications(BaselUniversityLibrary,Switzerland),http://www.bartoc.org

BasharatA.,AbroB.,ArpinarI.B.,&RasheedK.(2016):SemanticHadith:LeveragingLinkedDataOpportunitiesforIslamicKnowledge.In:LDOW2016-9thWorkshoponLinkedDataontheWeb,Montreal,Canada,12April2016,http://events.linkeddata.org/ldow2016/papers/LDOW2016_paper_06.pdf

BattenfeldI.,BeckmannI.,SchultzeJ.&TürkH.(2009):UnifyingArchaeologicalDatabasesusingTriples[ArcheoInf],pp.281-284,in:ProceedingsofCOINFO'09-FourthInternationalConferenceonCooperationandPromotionofInformationResourcesinScienceandTechnology,Beijing,China,IEEE,http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=5361890

BauerF.&KaltenböckM.(2012):LinkedOpenData:TheEssentials.AQuickStartGuideforDecisionMakers.REEP&SemanticWebCompany.Vienna:editionmono,http://www.semantic-web.at/LOD-TheEssentials.pdf

BechhoferS.,BuchanI.,DeRoureD.etal.(2011):Whylinkeddataisnotenoughforscientists,pp.300-307,in:E-Science’10-ProceedingsoftheIEEESixthInternationalConferenceone-Science,Brisbane,Australia,7-10December2010,http://eprints.soton.ac.uk/271587/5/research-objects-final.pdf

Beck,Anthony(2010):Digthenewbreed,PartIII–wrappingitallup.In:OpenKnowledgeBlog,11June2010,http://blog.okfn.org/2010/06/11/dig-the-new-breed-part-iii-wrapping-it-all-up/

Bedford,Denise(2014):UnderstandingandManagingTaxonomiesasEconomicGoodsandServices.In:ASIS&TBulletin,40(4):15-22,https://www.asist.org/publications/bulletin/apr-14/



Behkamal,Behshid(2014):MetricsDrivenFrameworkforLODQualityAssessment.ESWC2014-TheSemanticWeb:TrendsandChallenges.LectureNotesinComputerScience8465,pp.806-816,http://2014.eswc-conferences.org/sites/default/files/phdpaper_17.pdf

BenefielR.&SprenkleS.(2014):HerculaneumGraffitiProject.ISAWPaper7.4,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

Bénel,Aurélien(2015):SemioticIssuesandPerspectivesonModelingCulturalArtifactsRevisiting1970’sFrenchCriticismson‘Newarchaeologies’,pp.57-64,in:SWASH2016-1stWorkshoponSemanticWebforScientificHeritage,Portoroz,Slovenia,1June2015,http://ceur-ws.org/Vol-1364/sw4sh-2015.pdf

Bergman,MichaelK.(2014):ADecadeintheTrenchesoftheSemanticWeb.AI3weblog,16July2014,http://www.mkbergman.com/1771/a-decade-in-the-trenches-of-the-semantic-web/

BermanM.L.,MosternR.&SouthallH.(eds.,2016):PlacingNames:EnrichingandIntegratingGazetteers.Bloomington:IndianaUniversityPress(Series:TheSpatialHumanities),http://www.iupress.indiana.edu/product_info.php?products_id=808056

Berners-LeeT.,HendlerJ.&LassilaO.(2001):TheSemanticWeb.In:ScientificAmerican,May2001,http://www.sciam.com/2001/0501issue/0501berners-lee.html

Berners-Lee,Tim(1998–):DesignIssues,http://www.w3.org/DesignIssues/

Berners-Lee,Tim(2006):LinkedData,http://www.w3.org/DesignIssues/LinkedData.html

BikakisN.,TsinarakiC.,GioldasisN.,StavrakantonakisI.&ChristodoulakisS.(2013):TheXMLandSemanticWebWorlds:Technologies,InteroperabilityandIntegration.AsurveyoftheStateoftheArt.In:SemanticHyper/MultimediaAdaptation.StudiesinComputationalIntelligence,Vol.418,319-360,http://www.dblab.ntua.gr/~bikakis/papers/XMLSemanticWebSurvey.pdf

BindingC.&TudhopeD.(2016):ImprovingInteroperabilityusingVocabularyLinkedData.In:InternationalJournalonDigitalLibraries,17(1):5-21;acceptedmanuscript,http://hypermedia.research.glam.ac.uk/media/files/documents/2015-09-14/IJDL2015-binding-tudhope-P.docx

BindingC.,CharnoM.,JeffreyS.,MayK.&TudhopeD.(2015):TemplateBasedSemanticIntegration:FromLegacyArchaeologicalDatasetstoLinkedData.In:InternationalJournalonSemanticWebandInformationSystems,11(1),1-29.IGIGlobal,www.igi-global.com.Postedbypermissionofthepublisher.http://hypermedia.research.southwales.ac.uk/media/files/documents/2015-09-14/tudhope-paper_IJSWIS111.pdf

BindingC.,TudhopeD.,VlachidisA.etal.(2016):ARIADNE:AResearchInfrastructureforArchaeology.In:JournalonComputingandCulturalHeritage(forthcoming).

Binding,Ceri(2010):ImplementingarchaeologicaltimeperiodsusingCIDOCCRMandSKOS,pp.273-287,in:AroyoL.,AntoniouG.,HyvönenE.etal.(eds.):ESWC2010-TheSemanticWeb:ResearchandApplications.Springer(LNCS6088);preprint,http://www.researchgate.net/profile/Ceri_Binding/publication/225153456_Implementing_Archaeological_Time_Periods_Using_CIDOC_CRM_and_SKOS/links/0deec536b3f5384be7000000.pdf

Binding,Ceri(2014):5stardata–achievingthe5thstar.NKOS2014-13thEuropeanNetworkedKnowledgeOrganizationSystemsWorkshop,London,11September2014,https://at-web1.comp.glam.ac.uk/pages/research/hypermedia/nkos/nkos2014/programme.html

Bio2RDF:LinkedDatafortheLifeSciences,http://bio2rdf.org



BioPortal(USNationalCenterforBiomedicalOntology,providesaccesstoover300biological/bio-medicalvocabularies),https://bioportal.bioontology.org

BizerC.,HeathT.&Berners-LeeT.(2009):LinkedData-thestorysofar.In:InternationalJournalonSemanticWebandInformationSystems,5(3):1-22;preprint,http://eprints.soton.ac.uk/271285/1/bizer-heath-berners-lee-ijswis-linked-data.pdf

Bizer,Chris(2010):DataLinking,pp.34-43,in:GRDI2020-GlobalResearchDataInfrastructures:Towardsa10-yearvisionforglobalresearchdatainfrastructures,http://www.grdi2020.eu/Repository/FileScaricati/9a85ca56-c548-47e4-8b0e-86c3534ad21d.pdf

BlackwellC.&CraneG.(2009):Cyberinfrastructure,theScaifeDigitalLibraryandclassicsinadigitalage.In:DigitalHumanitiesQuarterly3(1),http://digitalhumanities.org/dhq/vol/3/1/000035/000035.html

BlackwellC.&SmithD.N.(2014):TheHomerMultitextandRDF-BasedIntegration.ISAWPaper7.5,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

Blumauer,Andreas(2013):TheLODcloudisdead,longlivethetrustedLODcloud.In:Semantic-Web.atweblog,(7June2013,http://blog.semantic-web.at/2013/06/07/the-lod-cloud-is-dead-long-live-the-trusted-lod-cloud/

Booth,David(2010):ResourceIdentityandSemanticExtensions:MakingSenseofAmbiguity.SemanticTechnologyConference,SanFrancisco,25-June-2010,http://dbooth.org/2010/ambiguity/paper.html

BozicB.&GordeaS.(2014):EnhancingtheLocalValueofThematicCulturalTourism.PATCHworkshop:TheFutureofExperiencingCulturalHeritage,partoftheIUI2014-Int.Conf.onIntelligentUserInterfaces,Haifa,Israel,24-27February2014,http://patch2014.files.wordpress.com/2012/07/submission-13-version-of-dec-24-10_08.pdf

BratkováE.&KučerováH.(2014):KnowledgeOrganizationSystemsandTheirTypology.In:RevueofLibrarianship,25(supplementum2):1-25,http://oldknihovna.nkp.cz/knihovna142_suppl/1402sup01.htm

BrewsterC.A.&O’HaraK.(2004):KnowledgeRepresentationwithOntologies:ThePresentandFuture.IEEEIntelligentSystems,January/February2004,https://www.inf.unibz.it/~franconi/papers/ieee-intelligent-systems-04.pdf

BrewsterC.A.&O’HaraK.(2007):Knowledgerepresentationwithontologies:Presentchallenges-Futurepossibilities.InternationalJournalofHuman-ComputerStudies,65(7):563-568,https://www.semanticscholar.org/paper/Knowledge-representation-with-ontologies-Present-Brewster-O%27Hara/69b7951abd61c63bb04636f7f51df8f2675d7417/pdf

BrewsterC.A.,IriaJ.,CiravegnaF.&WilksY.(2005):TheOntology:ChimaeraorPegasus.ProceedingsoftheDagstuhlSeminaronMachineLearningfortheSemanticWeb,February2005,http://eprints.aston.ac.uk/83/1/dagstuhl05.pdf

BritishMuseum-SemanticWebCollectionOnline,http://collection.britishmuseum.org

Buil-ArandaC.,HoganA.,UmbrichJ.&VandenbusscheP.Y.(2013):SPARQLWeb-QueryingInfrastructure:ReadyforAction?In:TheSemanticWeb-ISWC2013:12thInternationalSemanticWebConference,Sydney,21-25October2013,Proceedings,Part2:277-293,http://aidanhogan.com/docs/epmonitorISWC.pdf

Busch,JosephA.(2005):Makingthebusinesscasefortaxonomy(September27,2005),http://www.taxonomystrategies.com/presentations/BusinessCase.ppt



ByrneG.&GoddardL.(2010):Thestrongestlink:LibrariesandLinkedData.In:DLibMagazine,16(11/12),http://www.dlib.org/dlib/november10/byrne/11byrne.html

ByrneK.&KleinE.(2009):AutomaticExtractionofArchaeologicalEventsfromText.CAA2009-ComputerApplicationsinArchaeology,Williamsburg,Virginia,USA,http://homepages.inf.ed.ac.uk/kbyrne3/docs/byrneKleinCAA2009.pdf

Byrne,Kate(2006):TetheringCulturalDatawithRDF.InProceedingsofthe2006JenaUsersConference,Bristol,http://homepages.inf.ed.ac.uk/kbyrne3/docs/juc2006.pdf

Byrne,Kate(2008a):RelationalDatabasetoRDFTranslationintheCulturalHeritageDomain.SchoolofInformatics,UniversityofEdinburgh,May2008,http://homepages.inf.ed.ac.uk/s0233752/docs/rdb2rdfForCH.pdf

Byrne,Kate(2008b):HavingTriplets–HoldingCulturalDataasRDF.IACH2008-WorkshoponInformationAccesstoCulturalHeritage,ECDL2008,Aarhus,Denmark,18September2008,http://homepages.inf.ed.ac.uk/kbyrne3/docs/iach08kfb.pdf

Byrne,Kate(2009):PuttingHybridCulturalDataontheSemanticWeb.JournalofDigitalInformation(JoDI),10(6),http://homepages.inf.ed.ac.uk/kbyrne3/docs/jodi09kfb.pdf

CAASemanticSIG,https://groups.google.com/forum/#!forum/caa-semantic-sig

CacciottiR.&ValachJ.(2015):TheMONDISprojectSemanticWebandtheprotectionofhistoricbuildings,pp.307-313,in:ProceedingsofDigitalHeritage2015,Granada,Volume2,http://dx.doi.org/10.1109/DigitalHeritage.2015.7419512

CacciottiR.,BlaskoM.&ValachJ.(2014):Adiagnosticontologicalmodelfordamagestohistoricalconstruction.In:JournalofCulturalHeritage,16(1):40-48;preprint,https://www.academia.edu/10541233/A_diagnostic_ontological_model_for_damages_to_historical_constructions

CallouC.,BalyI.,GargominyO.&RiebE.(2011):NationalInventoryofNaturalHeritagewebsite:recent,historicalandarchaeologicaldata.In:TheSAAArchaeologicalRecord,11(1):37-40,http://alexandriaarchive.org/bonecommons/archive/files/kroeger_etal_icaz_saa_jan2011_f5bf7cdac2.pdf

CallouC.,BalyI.,MartinC.&LandaisE.(2009):BasededonnéesI2AF:InventairesarchéozoologiquesetarchéobotaniquesdeFrance.In:Archéopages,Issue26,Juillet2009,64-73,http://amenageurs.inrap.fr/userdata/c_bloc_file/13/13661/8449_fichier_pratiques-26.pdf

CallouC.,MichelF.,Faron-ZuckerC.,MartinC.&MontagnatJ.(2015):Towardsasharedreferencethesaurusforstudiesonhistoryofzoology,archaeozoologyandconservationbiology,pp.15-22,in:SWASH2016–1stWorkshoponSemanticWebforScientificHeritage,Portoroz,Slovenia,1June2015,http://ceur-ws.org/Vol-1364/sw4sh-2015.pdf

CalvaneseD.,LiuzzoP.,MoscaA.,RemesalJ.,RezkM.&RullG.(2016):Ontology-baseddataintegrationinEPNet:ProductionanddistributionoffoodduringtheRomanEmpire,pp.212–229,in:MiningtheHumanities:TechnologiesandApplications.EngineeringApplicationsofArtificialIntelligence,Volume51,May2016;preprint,https://www.semanticscholar.org/paper/Ontology-based-data-integration-in-EPNet-Calvanese-Liuzzo/3fad69e4e6a68f59b769042340c582e3d59d1f0b/pdf

CalvaneseD.,MoscaA.,RemesalJ.,RezkM.&RullG.(2015):A‘HistoricalCase’ofOntology-BaseddataAccess,pp.291-298,in:ProceedingsofDigitalHeritage2015,Granada,Volume2;preprint,http://ceipac.ub.edu/biblio/Data/A/0817.pdf



CarlisleP.K.,AvramidesI.,DalgityA.&MyersD.(2014):TheArchesHeritageInventoryandManagementSystem:AStandards-BasedApproachtotheManagementofCulturalHeritageInformation.PaperpresentedattheCIDOCConference:AccessandUnderstanding–NetworkingintheDigitalEra,Dresden,Germany,6-11September2014.http://archesproject.org/wp-content/uploads/2014/10/I-1_Carlisle_Dalgity_et-al_paper.pdf

CarverG.&LangM.(2013):Reflectionsontherockyroadtoe-archaeology,pp.224-236,in:CAA2012Southampton,VolumeI,AmsterdamUniversityPress,http://dare.uva.nl/cgi/arno/show.cgi?fid=516092

Carver,Geoff(2013):ArcheoInf,theCIDOC-CRMandSTELLAR:Workflow,Bottlenecks,andWheredoweGofromHere?,pp.498-508,in:CAA2012Southampton,VolumeII,AmsterdamUniversityPress,http://dare.uva.nl/cgi/arno/show.cgi?fid=545855

CasarosaV.,ManghiP.,MannocciA.,RiveroRuizE.&ZoppiF.(2014):AConceptualModelforInscriptions,pp.23-40,in:OrlandiS.etal.(eds.):InformationTechnologiesforEpigraphyandCulturalHeritage.ProceedingsoftheFirstEAGLEInternationalConference,Paris.Rome:SapienzaUniversitàEditrice,http://www.eagle-network.eu/wp-content/uploads/2015/01/Paris-Conference-Proceedings.pdf

CatalogueofLife,http://www.catalogueoflife.org

CATCHVocabularyandalignmentrepositorydemonstrator,http://www.cs.vu.nl/STITCH/repository/

CEIPAC-CentrefortheStudyofProvincialInterdependenceinClassicalAntiquity,UniversityofBarcelona,Spain,http://ceipac.ub.edu

CharlesV.&DevarenneC.(2014):EuropeanaenrichesitsdatawiththeAAT.EDMcasestudy,http://pro.europeana.eu/page/europeana-aat

CharlesV.,IsaacA.,FernieK.etal.(2013):AchievinginteroperabilitybetweentheCARAREschemaformonumentsandsitesandtheEuropeanaDataModel.ProceedingsofDC2013-InternationalConferenceonDublinCoreandMetadataApplications,http://dcevents.dublincore.org/IntConf/dc-2013/paper/view/171/171

CharnoM.,JeffreyS.,BindingC.,TudhopeD.&MayK.(2013):FromtheSlopeofEnlightenmenttothePlateauofProductivity:DevelopingLinkedDataattheADS,pp.216-223,in:CAA2012Southampton,VolumeI,AmsterdamUniversityPress,http://dare.uva.nl/cgi/arno/show.cgi?fid=516092

ChiarcosC.,LangM.&VerhagenP.(2015):IT-assistedExplorationofExcavationReports.UsingNaturalLanguageProcessingintheArchaeologicalResearchProcess,pp.87-93,in:CAA2015Siena,Proceedingsofthe43rdAnnualConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology.Oxford:Archaeopress,http://archaeopress.com/ArchaeopressShop/Public/download.asp?id={77DEDD4E-DE8F-43A4-B115-ABE0BB038DA7}

CIDOC(2012):StatementonLinkedDataidentifiersformuseumobjects.CIDOCAnnualGeneralMeeting,2012-06-13,Helsinki,http://network.icom.museum/fileadmin/user_upload/minisites/cidoc/PDF/StatementOnLinkedDataIdentifiersForMuseumObjects.pdf

CIDOCConceptualReferenceModel(CIDOCCRM),http://www.cidoc-crm.org

CIDOCCRM(2015):DefinitionoftheCIDOCConceptualReferenceModel.Version6.1,February2015,http://www.cidoc-crm.org/docs/cidoc_crm_version_6.1.pdf



CIDOCCRM:OverviewofCIDOCCRMextensions,http://www.ics.forth.gr/isl/CRMext/

CimianoP.,McCraeJ.,Rodriguez-DoncelV.etal.(2015):LinkedTerminology:ApplyingLinkedDataPrinciplestoTerminologicalResources,pp.504-517,in:ProceedingsofeLex2015-ElectronicLexicographyinthe21stcentury:LinkingLexicalData,HerstmonceuxCastle,Sussex,UK,11-13August2015,https://elex.link/elex2015/proceedings/eLex_2015_34_Cimiano+etal.pdf

CimianoP.,McCraeJ.P.&BuitelaarP.(2016):LexiconModelforOntologies.FinalCommunityGroupReport10May2016,https://www.w3.org/2016/05/ontolex/

CLAROS-ClassicalArtResearchOnlineServices,http://www.clarosnet.org

CLAROS:Data,http://data.clarosnet.org

Consens,MarianoP.(2013):ChallengesandOpportunitiesfortheOpenWebofLinkedData.WOD’2013-2ndInternationalWorkshoponOpenData,BNF,Paris(presentation),http://www-etis.ensea.fr/WOD2013/wp-content/uploads/2013/06/Consens-Challenges-and-Opportunities-for-the-Open-Web-of-Linked-Data.pdf

CorchoO.,Poveda-VillalónM.&Gómez-PérezA.(2015):OntologyEngineeringintheEraofLinkedData.In:ASIS&TBulletin41(4:Specialsection:LinkedDataandtheCharmofWeakSemantics),13-16,http://www.asis.org/Bulletin/Apr-15/Bulletin_AprMay2015.pdf

Coyle,Karen(2012):LinkedDataTools:ConnectingontheWeb.In:ALATechSource-LibraryTechnologyReports,48(4),http://www.alastore.ala.org/detail.aspx?ID=3845

Coyle,Karen(2013):DublinCoreusageinLOD.In:KCoyleweblog,9October2013,http://kcoyle.blogspot.co.at/2013/10/dublin-core-usage-in-lod.html

CreativeCommons(CC)licenses,https://creativecommons.org/licenses/

CrippsP.&MayK.(2004):ToOOornottoOO?RevelationsfromOntologicalModellingofanArchaeologicalInformationSystem.In:ProceedingsofComputerApplicationsandQuantitativeMethodsinArchaeology(CAA),Prato,Italy,13-17April2004,http://proceedings.caaconference.org/files/2004/08_Cripps_May_CAA_2004.pdf

CrippsP.,GreenhalghA.,FellowsD.,MayK.&RobinsonD.(2004):OntologicalModellingoftheworkoftheCentreforArchaeology.Technicalreport,EnglishHeritage-CentreforArchaeology,http://cidoc.ics.forth.gr/docs/Ontological_Modelling_Project_Report%_%20Sep2004.pdf

Cripps,Paul(2014):ColonisationofBritain.In:GeosemanticTechnologiesforArchaeologicalResearch(GSTAR)weblog,30May2014,http://gstar.archaeogeomancy.net/2014/05/colonisation-of-britain/

Cripps,Paul(2015):GeosemanticToolsforArchaeologicalResearch:GSTAR.PresentationatUSWAnnualPostgraduateResearchersPresentationDay,5May2015,http://de.slideshare.net/pauljcripps/uswpgr2015-cripps-gstar

CroftsN.,DoerrM.&Nyman(2011):CallforComments-LinkedOpenDataRecommendationforMuseums.CIDOCCRMwebsite,21March2011,http://www.cidoc-crm.org/URIs_and_Linked_Open_Data.html

CulturaItalia:Dati,http://dati.culturaitalia.it

CuyS.,GerthP.&FörtschR.(2016):ConnectingCulturalHeritageData:TheSyrianHeritageProjectintheITInfrastructureoftheGermanArchaeologicalInstitute,pp.251-258,in:CAA2015Siena-Proceedingsofthe43rdAnnualConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology.Oxford:Archaeopress,



http://archaeopress.com/ArchaeopressShop/Public/download.asp?id={77DEDD4E-DE8F-43A4-B115-ABE0BB038DA7}

D’Andrea,Andrea(2012):IncludingLinksinLinkedData:CIDOC-CRMandtheFourthT.Berners-LeeRule.VAST2012-13thInternationalSymposiumonVirtualReality,Archaeology,andCulturalHeritage.Brighton,UK,Nov.19-21,2012,https://www.academia.edu/4195188/Including_Links_in_Linked_Data_CIDOC-CRM_and_the_Fourth_T._Berners-Lee_Rule

D2RServer:AccessingdatabaseswithSPARQLandasLinkedData,http://d2rq.org/d2r-server

D2RQ-AccessingRelationalDatabasesasVirtualRDFGraphs,http://d2rq.org

DamovaM.&DannellsD.(2011):Reason-ableviewoflinkeddataforculturalheritage,pp.17-24,in:S3T-2011-ThirdInternationalConferenceonSoftware,ServicesandSemanticTechnologies.Springer(AISCvol.101);preprint,https://ontotext.com/documents/publications/2011/S3T-MuseumreasonableView_v7_cameraReady-30Jun.pdf

DamovaM.,DannéllsD.,EnacheR.,MatevaM.&RantaA.(2013):MultilingualaccesstoculturalheritagecontentontheSemanticWeb,pp.107–115,in:Proceedingsofthe7thWorkshoponLanguageTechnologyforCulturalHeritage,SocialSciences,andHumanities,Sofia,Bulgaria,8August2013,http://www.aclweb.org/anthology/W13-2715

DamovaM.,KiryakovA.,SimovK.&PetrovS.(2010):MappingtheCentralLODOntologiestoPROTONUpper-LevelOntology,pp.61-72,in:Proceedingsofthe5thInternationalConferenceonOntologyMatching,Shanghai,7November2010.CEUR-WS689,http://ceur-ws.org/Vol-689/om2010_Tpaper6.pdf

DANSlabs:EASYMetadataasLinkedOpenDataDemo,http://dans-labs.github.io/easy-lod/

DARIAH-DE(2013):RecommendationsforInterdisciplinaryInteroperability.Projectreport3.3.1,V1.0,15.02.2013,https://dev2.dariah.eu/wiki/download/attachments/14651583/R3.3.1.pdf?version=1&modificationDate=1366904278298&api=v2

DataHub(OpenKnowledgeFoundation),http://datahub.io

DBpedia(WikipediastructuredinformationoftenusedinLinkedDataprojects),http://dbpedia.org

deBoerV.&LeinengaJ.(2014):DiepereMaritiemeData.DANS.http://dx.doi.org/10.17026/dans-x8p-mc6a

deBoerV.,VanRossumM.,LeinengaJ.&HoekstraR.(2014):DutchShipsandSailorsLinkedData,pp.229-244,in:TheSemanticWeb-ISWC2014,Springer(LNCS8796);preprint,http://www.few.vu.nl/~vbr240/publications/deboer_iswc2014_dss_draft.pdf(datasets:http://datahub.io/dataset/dutch-ships-and-sailors)

deBoerV.,vanRossumM.,LeinengaJ.&HoekstraR.(2015):TheDutchShipsandSailorsProject.In:DHcommonsJournal,Issue1,July2015,http://dhcommons.org/journal/issue-1/dutch-ships-and-sailors-project

deBoerV.,WielemakerJ.,vanGentJ.etal.(2012):SupportingLinkedDataProductionforCulturalHeritageInstitutes:TheAmsterdamMuseumCaseStudy.Proceedingsofthe9thExtendedSemanticWebConference(ESWC2012),TPDLconference.Heraklion,Greece.27-31May2012,http://www.few.vu.nl/~vbr240/publications/eswc2012supporting.pdf



deBoerV.,WielemakerJ.,vanGentJ.etal.(2013):AmsterdamMuseumLinkedOpenData.In:SemanticWebJournal,4(3):237-243,http://www.semantic-web-journal.net/sites/default/files/swj293_2.pdf

DeBoer,Victor(2015):LinkedDataforDigitalHistory,pp.5-6,in:SWASH2016-1stWorkshoponSemanticWebforScientificHeritage,Portoroz,Slovenia,1June2015,http://ceur-ws.org/Vol-1364/sw4sh-2015.pdf

DeclerckT.,Wandl-VogtE.&MörthK.(2015):TowardsaPanEuropeanLexicographybyMeansofLinked(Open)Data,pp.342-355,in:ProceedingsofeLex2015-ElectronicLexicographyinthe21stcentury:LinkingLexicalData,HerstmonceuxCastle,Sussex,UK,11-13August2015,https://elex.link/elex2015/proceedings/eLex_2015_22_Declerck+etal.pdf

DiGiorgioS.,FelicettiA.,MartiniP.&MasciE.(2016):Dati.CulturaItalia:aUseCaseofPublishingLinkedOpenDataBasedonCIDOC-CRM,pp.44-54,in:Ronzino,Paola(ed.):Extending,MappingandFocusingtheCRM.ProceedingsoftheEMF-CRMworkshop,Poznan,Poland,17September2015,http://ceur-ws.org/Vol-1656/paper4.pdf

DigitalAtlasoftheRomanEmpire(DepartmentofArchaeologyandAncientHistory,LundUniversity,Sweden),http://dare.ht.lu.se

DigitalCollaboratoryforCulturalDendrochronology-DCCD,http://dendro.dans.knaw.nl;projectwebsite:http://vkc.library.uu.nl/vkc/dendrochronology/

DigitalObjectIdentifierSystem,http://www.doi.org

DoddsL.&DavisI.(2012):LinkedDataPatterns.Apatterncatalogueformodelling,publishing,andconsumingLinkedData(version2012-05-31),http://patterns.dataincubator.org/book/

Dodds,Leighetal.(2010):QualityIndicatorsforLinkedDataDatasets.DiscussiononSemantic-Overflow,24.06.-13.07.2010,http://answers.semanticweb.com/questions/1072/quality-indicators-for-linked-data-datasets

DoerrM.&HiebelG.(2013):CRMgeo:LinkingtheCIDOCCRMtoGeoSPARQLthroughaspatiotemporalrefinement.ICS-FORTH/TR-435,April2013,https://www.ics.forth.gr/tech-reports/2013/2013.TR435_CRMgeo_CIDOC_CRM_GeoSPARQL.pdf

DoerrM.&OldmanD.(2013):TheCostsofCulturalHeritageDataServices:TheCIDOCCRMorAggregatorformats?DominicOldmanweblog,13June2013,http://www.oldman.me.uk/blog/costsofculturalheritage/

DoerrM.,BekiariC.,KritsotakiA.,HiebelG.&TheodoridouM.(2014a):ModellingScientificActivities:Proposalforaglobalschemaforintegratingmetadataaboutscientificobservation.PaperpresentedattheCIDOC2014Conference,6th-11thSept.2014,Dresden/Germany,http://www.cidoc2014.de/images/sampledata/cidoc/papers/E-2_Bekiari_paper.pdf

DoerrM.,deJongG.,KonsolakiK.,NortonB.,OldmanD.,TheodoridouM.&WikmanT.(2014b):TheSYNERGYReferenceModelofDataProvisionandAggregation.Draft,June2014,http://www.cidoc-crm.org/docs/SRM_v0.1.pdf

DoerrM.,KritsotakiA.&Boutsika,A.(2011):Factualargumentation-acoremodelforassertionsmaking.In:JournalonComputingandCulturalHeritage(JOCCH),3(3),http://dl.acm.org/citation.cfm?id=1921615

DoerrM.,SchallerK.&TheodoridouM.(2004):Integrationofcomplementaryarchaeologicalsources.In:NiccolucciF.(ed.):Proceedingsofthe32ndComputerApplicationsandQuantitative



MethodsinArchaeologyConference,http://proceedings.caaconference.org/files/2004/09_Doerr_et_al_CAA_2004.pdf

DoerrM.,TheodoridouM.,AspöckE.&MasurA.(2016):MappingArchaeologicalDatabasestoCIDOCCRM,pp.443-451,in:CAA-2015-43rdConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology(Siena,April2015).Oxford:Archaeopress,http://archaeopress.com/ArchaeopressShop/Public/download.asp?id={77DEDD4E-DE8F-43A4-B115-ABE0BB038DA7}

Doerr,Martin(2010):TechnologicalChoicesoftheResearchSpaceProject.Researchspace.org,August2010,http://www.researchspace.org/researchspace-concepts/technological-choices-of-the-researchspace-project

DuanS.,KementsietsidisA.,SrinivasK.&UdreaO.(2011):Applesandoranges:acomparisonofRDFbenchmarksandrealRDFdatasets.In:SIGMOD’11,Athens,Greece,12–16June2011,conferenceproceedings,pp.145–156,http://researcher.ibm.com/researcher/files/us-sduan/sigmod2011_RDF_benchmark_duan.pdf

DublinCoreMetadataElementSet,Version1.1,2012-06-14,http://dublincore.org/documents/dces/

DublinCoreMetadataInitiative(DCMI)MetadataTerms,http://dublincore.org/documents/dcmi-terms/

DunsireG.,HarperC.,HillmannD.&PhippsJ.(2012):LinkedDataVocabularyManagement:InfrastructureSupport,DataIntegration,andInteroperability.In:InformationStandardsQuarterly,24(2/3):4-13,http://www.niso.org/publications/isq/2012/v24no2-3/dunsire/

DutchShipsandSailors(ClarinIVproject,4/2013-3/2014),http://dutchshipsandsailors.nl

EAGLE-EuropeanaNetworkofAncientGreekandLatinEpigraphy(EU,ICT-PSP,4/2013-3/2016),http://www.eagle-network.eu

EAGLE(2015):EAGLEMetadataModelSpecification–Secondrelease.ProjectdeliverableD3.1.2,V1.1,26January2015,http://www.eagle-network.eu/wp-content/uploads/2013/06/EAGLE_D3.1_EAGLE-metadata-model-specification_v1.1.pdf

EAGLEvocabularies(Material,Typeofinscription,Executiontechnique,Objecttype,Decoration,Datingcriteria,Stateofpreservation),http://www.eagle-network.eu/resources/vocabularies/

EckkrammerF.,FeldbacherR.&EckkrammerT.(2011):CIDOCCRMinDataManagementandDataSharing.DataSharingbetweenDifferentDatabases,pp.80-85,in:CAA-2008.36thAnnualConferenceofComputerApplicationsandQuantitativeMethodsinArchaeology,Budapest;http://proceedings.caaconference.org/files/2008/CD19_Eckkrammer_et_al_CAA2008.pdf

EdelsteinJ.,GallaL.,Li-MadeoC.,MardenJ.RhonemusA.&WhyselN.(2013a):LinkedOpenDataforCulturalHeritage:EvolutionofanInformationTechnology.NewYork:PrattInstitute,Spring2013,http://www.whysel.com/papers/LIS670-Linked-Open-Data-for-Cultural-Heritage.pdf

EdelsteinJ.,Li-MadeoC.,MardenJ.&WhyselN.(2013b):LinkedOpenDataforCulturalHeritage:evolutionofaninformationtechnology,pp.107-112,in:SIGDOC’13-Proceedingsofthe31stACMInternationalConferenceonDesignofCommunication;preprint,http://academiccommons.columbia.edu/catalog/ac:168445

ElliottT.&GilliesS.(2009):DigitalGeographyandClassics.In:DigitalHumanitiesQuarterly,3(1),http://www.digitalhumanities.org/dhq/vol/3/1/000031.html



ElliottT.&JonesC.(2014):MovingtheAncientWorldOnlineForward.ISAWPaper7.6,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

ElliottT.,HeathS.&MuccigrossoJ.(2012):ReportontheLinkedAncientWorldDataInstitute.In:ISQ-InformationStandardsQuarterly,24(2/3):43-45,http://dx.doi.org/10.3789/isqv24n2-3.2012.08

ElliottT.,HeathS.&MuccigrossoJ.(2014):PrologueandIntroduction.ISAWPaper7.1,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

ElliottT.,HeathS.&MuccigrossoJ.(eds.,2014):CurrentPracticeinLinkedOpenDatafortheAncientWorld.InstitutefortheStudyoftheAncientWorld,NewYorkUniversity.ISAWPapers7,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

EncodedArchivalDescription,http://www.loc.gov/ead/

EncyclopediaofLife(EOL),http://www.eol.org

EnglishHeritagePlaces,DataHubinformation,http://datahub.io/dataset/englishheritage_places

Entjes,JeroenA.(2015):LinkingMaritimeDatasetstoDutchShipsandSailorsCloud-CasestudiesonArchangelvaartandElbing.Masterthesisproject,https://videbo.files.wordpress.com/2015/08/jeroen_entjes_final_thesis.pdf

EnvironmentOntology,https://bioportal.bioontology.org/ontologies/ENVO

EpiDoc:EpigraphicDocumentsinTEIXML,http://epidoc.sf.net

EpigraphicDatabaseHeidelberg,http://edh-www.adw.uni-heidelberg.de

EPNet-ProductionandDistributionofFoodduringtheRomanEmpire:EconomicandPoliticalDynamics(ERCAdvancedGrantproject,3/2014-2/2019),http://www.roman-ep.net

EpureE.V.,Martín-RodillaP.,HugC.,DeneckèreR.&SanilesiC.(2015):AutomaticProcessModelDiscoveryfromTextualMethodologies:AnArchaeologyCaseStudy.ProceedingsofRCIS2015-NinthIEEEInternationalConferenceonResearchChallengesinInformationScience,Athens,Greece,May2015,https://hal-paris1.archives-ouvertes.fr/hal-01149742/document

Erdman-ThomsenH.,Pareja-LoraA.&NistrupMadsenB.(2016):TermBasesandLinguisticLinkedOpenData.TKE2016-12thInternationalconferenceonTerminologyandKnowledgeEngineering.CopenhagenBusinessSchool,http://openarchive.cbs.dk/handle/10398/9323

ErmilovI.,LehmannJ.,MartinM.&AuerS.(2016):LODStats:TheDataWebCensusDataset.ISWC2016-15thInternationalSemanticWebConference,Kobe,Japan,17-21October2016;preprint,https://svn.aksw.org/papers/2016/ISWC_LODStats_Resource_Description/public.pdf

ErpM.,OomenJ.,SegersR.etal.(2011):Automaticheritagemetadataenrichmentwithhistoricevents.MuseumsandtheWeb2011,6-9April2011,Philadelphia,http://www.museumsandtheweb.com/mw2011/papers/automatic_heritage_metadata_enrichment_with_hi

ErxlebenF.,GüntherM.,KrötzschM.,MendezJ.&VrandeciD.(2014):IntroducingWikidatatotheLinkedDataWeb.ISWC2014-13thInternationalSemanticWebConference,RivadelGarda,Italy,http://korrekt.org/papers/Wikidata-RDF-export-2014.pdf

EUCLID-EducationalCurriculumfortheUsageofLinkedData,http://euclid-project.eu

EuropeanCoinFindNetwork(ECFN),http://www.ecfn.fundmuenzen.eu



EuropeanCommission,JoinupPortal-Shareandreuseinteroperabilitysolutionsforpublicadministrations,http://joinup.ec.europa.eu

EuropeanLanguageSocialScienceThesaurus(ELSST),http://elsst.ukdataservice.ac.uk

EuropeanNetworkofe-Lexicography-ENeL(EU,COSTAction,10/2013-10/2017,http://www.elexicography.eu

EuropeanPersistentIdentifierConsortium(EPIC),http://www.pidconsortium.eu

EuropeanaCloudproject(02/2013-01/2015,CIP-ICT-PSPBestPracticeNetwork,http://pro.europeana.eu/web/europeana-cloud

EuropeanaDataModel(EDM),http://pro.europeana.eu/edm-documentation

EuropeanaLinkedData,http://labs.europeana.eu/api/linked-open-data/introduction/

EuropeanaTechTaskForceonaMultilingualandSemanticEnrichmentStrategy:finalreport,7April2014,http://pro.europeana.eu/documents/468623/8b75b054-712e-432b-a0f7-761898e6f60e

EuropeanaConnect(EU,eContent+project,5/2009-10/2011),http://www.europeanaconnect.eu

FaBiO-FRBR-alignedBibliographicOntology,http://vocab.ox.ac.uk/fabio

Faron-ZuckerC.,PajónLeyraI.,PoulidaK.&TettamanziA.(2016):SemanticCategorizationofSegmentsofAncientandMediaevalZoologicalTexts,pp.59-68,SWASH2016-2ndWorkshoponSemanticWebforScientificHeritage,Heraklion,Greece,30May2016,http://ceur-ws.org/Vol-1595/paper7.pdf

FelicettiA.&LorenziniM.(2011):Metadataandtoolsforintegrationandpreservationofculturalheritage3Dinformation.23rdInternationalCIPASymposium,Prague,CzechRepublic,12-16September2011,http://cipa.icomos.org/fileadmin/template/doc/PRAGUE/051.pdf

FelicettiA.,GalluccioI.,LuddiC.,MancinelliM.L.,ScarselliT.&MadonnaA.D.(2016):IntegratingTerminologicalToolsandSemanticArchaeologicalInformation:theICCDRASchemaandThesaurus,pp.28-43,in:Ronzino,Paola(ed.):Extending,MappingandFocusingtheCRM.ProceedingsoftheEMF-CRMworkshop,Poznan,Poland,17September2015,http://ceur-ws.org/Vol-1656/paper3.pdf

FelicettiA.,GerthP.,MeghiniC.&TheodoridouM.(2016):IntegratingHeterogeneousCoinDatasetsintheContextofArchaeologicalResearch,pp.13-27,in:Ronzino,Paola(ed.):Extending,MappingandFocusingtheCRM.ProceedingsoftheEMF-CRMworkshop,Poznan,Poland,17September2015,http://ceur-ws.org/Vol-1656/paper2.pdf

FelicettiA.,MuranoF.,RonzinoP.&NiccolucciF.(2016):CIDOCCRMandEpigraphy:aHermeneuticChallenge,pp.55-68,in:Ronzino,Paola(ed.):Extending,MappingandFocusingtheCRM.ProceedingsoftheEMF-CRMworkshop,Poznan,Poland,17September2015,http://ceur-ws.org/Vol-1656/paper5.pdf

FelicettiA.,ScarselliT.,MancinelliM.L.&NiccolucciF.(2013):MappingICCDArchaeologicalDatatoCIDOC-CRM:theRASchema.In:AlexievV.etal.(eds.):PracticalExperienceswithCIDOCCRManditsExtensions(CRMEX2013)Workshop,17thInternationalConferenceonTheoryandPracticeofDigitalLibraries(TPDL2013),Valetta,Malta,26September2013,http://ceur-ws.org/Vol-1117/paper2.pdf

Felicetti,Achille(2012):Digitalcollectionsofsemanticallyannotatedculturalheritagetexts.In:UncommonCulture,Vol.3,no.5/6(2012):61-64,http://journals.uic.edu/ojs/index.php/UC/article/view/4719/3682



FelleA.E.&RoccoA.(eds.,2016):OfftheBeatenTrack.EpigraphyattheBorders.ProceedingsoftheVIEAGLEInternationalEvent,24-25September2015,Bari,Italy.Oxford:Archaeopress,http://www.archaeopress.com/ArchaeopressShop/Public/download.asp?id={E7B2AAC6-9986-4C41-9842-6AA93BE7ACD9}

FerraraA.,NikolovA.&ScharffeF.(2011):DataLinkingfortheSemanticWeb.In:InternationalJournalontheSemanticWebinInformationSystems,7(3):46-76,manuscript,http://people.kmi.open.ac.uk/andriy/data_linking_for_the_semantic_web.pdf(paper:http://www.igi-global.com/article/data-linking-semantic-web/62562

FerroN.,MunnellyG.,HampsonC.&ConlanO.(2013):FosteringInteractionwithCulturalHeritageMaterialviaAnnotations:TheFAST-CATWay.Proceedingsofthe9thItalianResearchConferenceonDigitalLibraries(IRCDL2013),CCISVol.385,http://www.tara.tcd.ie/xmlui/bitstream/handle/2262/67966/fast-cat-IRCDL2013.v2%20copy.pdf;jsessionid=B36C67BA30EC9A76F83C8BDE7A6A03DC?sequence=1

Finto-Finnishthesaurusandontologyservice,http://finto.fi/en/

FOAF-Friend-of-a-Friend,http://xmlns.com/foaf/spec/

ForumonInformationStandardsinHeritage(FISH):http://heritage-standards.org.uk/fish-vocabularies/

Fossilworks,http://fossilworks.org

FreeYourMetadataproject(iMinds/GhentUniversityandMaSTIC/UniversitéLibredeBruxelles)http://freeyourmetadata.org

FreitasA.,CurryE.,OliveiraJ.G.&O’RiainS.(2012):QueryingHeterogeneousDatasetsontheLinkedDataWeb:Challenges,Approaches,andTrends.IEEEInternetComputing,16(1),January/February201224-33,http://www.edwardcurry.org/publications/freitas_IC_12.pdf

FürberC.&HeppM.(2010a):UsingSemanticWebResourcesforDataQualityManagement,pp.211-225,in:KnowledgeEngineeringandManagementbytheMasses.Springer:LectureNotesinComputerScienceVolume6317;preprint,http://www.fuerber.com/publications/Fuerber-Hepp-Using_Semantic_Web_Resources_for_Data_Quality_Management.pdf

FürberC.&HeppM.(2010b):UsingSPARQLandSPINforDataQualityManagementontheSemanticWeb.In:BusinessInformationSystems.LectureNotesinBusinessInformationProcessing,Vol.47:35-46,http://www.heppnetz.de/files/fuerber-hepp-sparql-spin-dqm.pdf

FürberC.&HeppM.(2011a):SWIQA-ASemanticWebInformationQualityAssessmentFramework.ECIS2011-EuropeanConferenceonInformationSystems,Proceedings,paper76,http://aisel.aisnet.org/cgi/viewcontent.cgi?article=1075&context=ecis2011

FürberC.&HeppM.(2011b):DataQualityManagementVocabulary.V1.0,9October2011,http://semwebquality.org/dqm-vocabulary/v1/dqm

FürberC.,HeppM.&WischnewskiM.(2011):DataQualityConstraintsLibrary.V1.1,28March2011,http://semwebquality.org/ontologies/dq-constraints

GBIF(2011):RecommendationsfortheUseofKnowledgeOrganisationSystemsbyGBIF.Releasedon4February2011.Copenhagen:GlobalBiodiversityInformationFacility,http://www.gbif.org/resource/80656

GeigerC.P.&vonLuckeJ.(2012):OpenGovernmentand(Linked)(Open)(Government)(Data).Freeaccessibledataofthepublicsectorinthecontextofopengovernment.In:JeDEM-eJournalof



eDemocracyandOpenGovernment,4(2):265-278,http://www.jedem.org/index.php/jedem/article/download/143/115

GEMET-GeneralMultilingualEnvironmentalThesaurus(EIONET/EuropeanEnvironmentAgency),http://www.eionet.europa.eu/gemet/

GeologicalSurveyofIreland,http://www.gsi.ie

GeoNames,http://www.geonames.org

GeoSpeciesontology,https://bioportal.bioontology.org/ontologies/GEOSPECIES

GermanNationalLibrary:LinkedDataService,http://dnb.de/EN/lds

GerthP.,SchmidleW.&CuyS.(2016a):SculpturesintheSemanticWeb.PresentationatCAA2016-44thComputerApplicationsandQuantitativeMethodsinArchaeologyConference,Oslo,Norway,30March2016,http://www.slideshare.net/ariadnenetwork/sculptures-in-the-semantic-web-65237911

GerthP.,SchmidleW.&CuyS.(2016b):SculpturesintheSemanticWeb.In:ProceedingsofCAA2016-44thComputerApplicationsandQuantitativeMethodsinArchaeologyConference,Oslo,Norway,29March-2April2016(paperforthcoming).

Geser,Guntram(2003):ACulturalHeritageSemanticWebExample&Primer,pp.26-36,in:DigiCULTThematicIssue3:TowardsaSemanticWebforHeritageResources.Salzburg,May2003,http://www.digicult.info/pages/Themiss.php

Geser,Guntram(2004):Assessingthereadinessofsmallheritageinstitutionsfore-culturetechnologies,pp.8-13,in:DigiCULT.Infoe-Journal,Issue9,November2004,http://www.digicult.info/downloads/digicult_info_9.pdf

Geser,Guntram(2009):STERNATechnologyWatchReport.AReportonSemanticApproachesforIncludingDigitalCulturalandBio-HeritageResourcesintheEuropeanDigitalLibraryInitiative.Salzburg,January2009,http://www.sterna-net.eu/images/stories/documents/sterna_del.6.5_technology-watch_full-report_20081210.pdf

Geser,Guntrametal.(2003):TowardsaSemanticWebforHeritageResources.DigiCULTThematicIssue3,May2003,http://www.digicult.info/downloads/thematic_issue_3_low.pdf

GettyVocabulariesasLinkedOpenData,http://www.getty.edu/research/tools/vocabularies/lod/

GettyVocabularies:LOD,http://vocab.getty.edu

GoddardL.&ByrneG.(2010):LinkedDatatools:SemanticWebforthemasses.In:FirstMonday,15(11),http://firstmonday.org/ojs/index.php/fm/article/view/3120/2633

GoldenP.&ShawR.(2015):Periodassertionasnanopublication.ThePeriodOperiodgazetteer.In:Semantics,Analytics,Visualisation:EnhancingScholarlyData.WorkshopCo-LocatedwithWWW’15-24thInternationalWorldWideWebConference,Florence,Italy.http://cs.unibo.it/save-sd/2015/papers/html/golden-savesd2015.html

GoldenP.&ShawR.(2016):Nanopublicationbeyondthesciences:thePeriodOperiodgazetteer.In:PeerJComputerScience2:e44,https://peerj.com/articles/cs-44/

GolubK.&TudhopeD.(2009):TerminologyRegistryScopingStudy(TRSS):Finalreport,3July2009,http://www.jisc.ac.uk/media/documents/programmes/sharedservices/trss-report-final.pdf



GolubK.,TudhopeD.,ZengM.L.&ŽumerM.(2014):Terminologyregistriesforknowledgeorganizationsystems:Functionality,use,andattributes.In:JournaloftheAssociationforInformationScience&Technology,65(9):1901-16,http://dx.doi.org/doi:10.1002/asi.23090

GoodB.M.&WilkinsonM.D.(2006):TheLifeSciencesSemanticWebisFullofCreeps!BriefingsinBioinformatics20067(3):275-286,http://bib.oxfordjournals.org/cgi/content/full/7/3/275?ck=nck#T1

GörzG.&ScholzM.(2012):WissKI:AVirtualResearchEnvironmentforCulturalHeritage.In:DeRaedt,Lucetal.(eds.):ECAI2012-20thEuropeanConferenceonArtificialIntelligence,Montpellier27-31August2012.Amsterdam:IOSPress;preprint,http://wwwdh.cs.fau.de/IMMD8/staff/Goerz/ecai2012.pdf

GracyK.&LambertF.(2014):Who’sreadytosurfthenextwave?Astudyofperceivedchallengestoimplementingnewandrevisedstandardsforarchivaldescription.In:TheAmericanArchivist,77(1):96-132,http://americanarchivist.org/doi/abs/10.17723/aarc.77.1.b241071w5r252612

Gracy,KarenF.(2015):Archivaldescriptionandlinkeddata:apreliminarystudyofopportunitiesandimplementationchallenges.In:ArchivalScience,15(3):239-294,http://link.springer.com/article/10.1007/s10502-014-9216-2

GrassiM.,MorbidoniC.,NucciM.etal.(2013):Pundit:AugmentingWebContentswithSemantics.LiteraryandLinguistingComputing,Vol.28,No.4,http://dm2e.eu/files/Graasi-et-al.-2013-Pundit-augmenting-web-contents-with-semantics.pdf

Gros,Jean-Sébastien(2016):Atλaς,aGazetteerLinkingArchaeologicalCollections,pp.19-24,in:SWASH2016-2ndWorkshoponSemanticWebforScientificHeritage,Heraklion,Greece,30May2016,http://ceur-ws.org/Vol-1595/paper2.pdf

GruberE.&SmithT.J.(2014):LinkedOpenGreekPottery,pp.205-214,in:CAA2014Paris-Proceedingsofthe42ndAnnualConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology,Archaeopress;preprint,https://www.academia.edu/9739936/Linked_Open_Greek_Pottery

GruberE.,BransbourgG.,HeathS.&MeadowsA.(2013):LinkingRomanCoins:CurrentWorkattheAmericanNumismaticSociety,pp.249-258,in:CAA2012Southampton,VolumeI.AmsterdamUniversityPress;preprint,https://www.academia.edu/6604014/Linking_Roman_Coins_Current_Work_at_the_American_Numismatic_Society

GruberE.,GondekR.&SmithT.J.(2015):CAA2015Siena,Roundtable–LinkedOpenDataAppliedtoPotteryDatabases,1April2015,http://2015.caaconference.org/program/roundtables/rt3/

Gruber,Ethan(2016):LODforNumismaticLAMIntegration.PresentationatCAA2016Oslo,session“LinkedPasts:ConnectingIslandsofContent”,30March2016,http://de.slideshare.net/ewg118/lod-for-numismatic-lam-integration

GruntgensM.&SchradeT.(2016):DatarepositoriesintheHumanitiesandtheSemanticWeb:modelling,linking,visualising,pp.53-64,in:ProceedingsofWHiSe2016-1stWorkshoponHumanitiesintheSemanticWeb,Anissaras,Greece,29May2016,http://ceur-ws.org/Vol-1608/paper-07.pdf

GueguenG.,MarquesdaFonsecaV.M.,PittiD.V.&SibilledeGrimoüardC.(2013):TowardanInternationalConceptualModelforArchivalDescription:APreliminaryReportfromthe



InternationalCouncilonArchives’ExpertsGrouponArchivalDescription.In:TheAmericanArchivist,76(2):566-582,http://www.ica.org/sites/default/files/EGAD_English.pdf

HADOC–Harmonisationdelaproductiondesdonnéesculturellesprogramme(MinistèredelaCultureetdelaCommunication,France),http://www.culturecommunication.gouv.fr/Ressources/Harmonisation-des-donnees-culturelles

HaferL.W.&KirkpatrickA.E.(2009):Assessingopensourcesoftwareasascholarlycontribution.In:CommunicationsoftheACM,52(12:,126-129,http://dl.acm.org/citation.cfm?id=1610285&CFID=500091250&CFTOKEN=70398928

Hafford,WilliamB.(2014):LinkedOpenDataandtheUroftheChaldeesProject.ISAWPaper7.7,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

HalbW.&HausenblasM.(2008):select*where{:I:trust:you}.HowtoTrustInterlinkedMultimediaData.ProceedingsoftheInternationalWorkshoponInteractingwithMultmediaContentintheSocialSemanticWeb(IMC-SSW2008),Koblenz,Germany,3December2008,http://ceur-ws.org/Vol-417/paper6.pdf

HannemannJ.&KettJ.(2010):LinkedDataforLibraries.IFLA2010–WorldLibraryandInformationCongress,Gothenburg,Sweden,10-15August2010,http://conference.ifla.org/past-wlic/2010/149-hannemann-en.pdf

Harpring,Patricia(2014):LinkedOpenDataintheCulturalHeritageWorld:IssuesforInformationCreatorsandUsers.CLIR–CouncilonLibraryandInformationResourcesweblog,20March2014,http://connect.clir.org/blogs/patricia-harpring/2014/03/20/linked-open-data-in-the-cultural-heritage-world-issues-for-information-creators-and-users

Harpring,Patricia(2016):Art&ArchitectureThesaurus.IntroductionandOverview.GettyVocabulayProgram,http://www.getty.edu/research/tools/vocabularies/aat_in_depth.pdf

Hart,Glen(2009):Linkingtothepast,geographicallyspeaking:TheLinkedDataWeb&HistoricalGIS,http://www.ordnancesurvey.co.uk/oswebsite/partnerships/research/publications/docs/2009/Linking_to_the_Past_GeoS.pdf

HaslhoferB.,MomeniE.,Gay,M.&SimonR.(2010):AugmentingEuropeanaContentwithLinkedDataResources.Proceedingsofthe6thInternationalConferenceonSemanticSystems(I-Semantics),Graz,Austria,1-3September2010,http://eprints.cs.univie.ac.at/26/1/ldtc2010_haslhofer_et_al_cr2.pdf

HasnainA.,SanaeZainabS.,KamdarM.R.etal.(2015):ARoadmapfornavigatingtheLifeSciencesLinkedOpenDataCloud,pp.97-112,in:SemanticTechnology-4thJointInternationalConference,JIST2014,ChiangMai,Thailand,9-11November2014.Springer(LNCS8943);preprint,http://maulik-kamdar.com/wp-content/uploads/2014/10/JIST2014.pdf

HausteinS.&PleumannJ.(2002):IsParticipationintheSemanticWebTooDifficult?,pp.448-453,in:HorrocksI.&HendlerJ.(eds.):TheSemanticWeb-ISWC2002.Berlin:Springer,http://sfb876.tu-dortmund.de/PublicPublicationFiles/haustein_pleumann_2002b.pdf

HeathT.&BizerC.(2011):LinkedData:EvolvingtheWebintoaGlobalDataSpace(1stedition).SynthesisLecturesontheSemanticWeb:TheoryandTechnology,1:1,1-136.Morgan&Claypool.Online:http://linkeddatabook.com/editions/1.0/

Heath,Sebastian(2014):ISAWPapers:TowardsaJournalasLinkedOpenData.ISAWPaper7.8,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/



Heath,Tom(2009):LinkedData?WebofData?SemanticWeb?WTF?,http://tomheath.com/blog/2009/03/linked-data-web-of-data-semantic-web-wtf/

Heath,Tom(2010):WhyCarrytheCostofLinkedData?TomHeathweblog,16June2010,http://tomheath.com/blog/2010/06/why-carry-the-cost-of-linked-data/

HennickeS.,MarliesOlenskyM.,deBoerV.etal.(2011):Adatamodelforcross-domaindatarepresentation:TheEuropeanaDataModelinthecaseofarchivalandmuseumdata.Proceedingsofthe12thInternationalSymposiumonInformationScience,(ISI2011).Hildesheim,Germany,March9-112011.http://www.few.vu.nl/~AI.Isaac/publications.php

Hepp,Martin(2007):PossibleOntologies.HowRealityConstrainstheDevelopmentofRelevantOntologies.In:IEEEInternetComputing,11(1):90-96,http://www.heppnetz.de/files/IEEE-IC-PossibleOntologies-published.pdf

HeritageData-LinkedDataVocabulariesforCulturalHeritage,http://www.heritagedata.org

Hirst,Tony(2010):Commentsto“WhyCarrytheCostofLinkedData?”.TomHeathweblog,16June2010,http://tomheath.com/blog/2010/06/why-carry-the-cost-of-linked-data/

Hodge,Gail(2014):GovernmentKnowledgeOrganizationSystems:ValuingaPublicGood,pp.23-29,in:ASIS&TBulletin,40(4),April/May2014,https://www.asist.org/publications/bulletin/apr-14/

HoekstraR.,Meroño-PeñuelaA.,DentlerK.,RijpmaA.,ZijdemanR.&ZandhuisI.(2016):AnecosystemforLinkedHumanitiesData,pp.85-96,In:ProceedingsofWHiSe2016-1stWorkshoponHumanitiesintheSemanticWeb,Anissaras,Greece,29May2016,http://ceur-ws.org/Vol-1608/paper-11.pdf

HoganA.&GutierrezC.(2014):PathstowardstheSustainableConsumptionofSemanticDataontheWeb.AMW2014-8thAlbertoMendelzonWorkshoponFoundationsofDataManagement,CartagenadeIndias,Colombia,4-6June2014,CEURWorkshopProceedings,http://ceur-ws.org/Vol-1189/paper_7.pdforhttp://ciws.cl/media/pdf/amw_2014_hogan.pdf

HoganA.,HarthA.,PassantA.,DeckerS.&PolleresA.(2010):Weavingthepedanticweb.In:LDOW2010-3rdInternationalWorkshoponLinkedDataontheWeb,Raleigh,USA,27April2010,http://events.linkeddata.org/ldow2010/papers/ldow2010_paper04.pdf

HoganA.,UmbrichJ.,HarthA.etal.(2012):AnempiricalsurveyofLinkedDataconformance.In:WebSemantics:Science,ServicesandAgentsontheWorldWideWeb,SpecialIssueon‘DealingwiththeMessinessoftheWebofData’,volume14,July2012,http://aidanhogan.com//docs/ldstudy12.pdf

HolmenJ.&OreC.(2010):DeducingEventChronologyinaCulturalHeritageDocumentationSystem.In:Proceedingsofthe38thComputerApplicationsandQuantitativeMethodsinArchaeology(CAA)Conference,Williamsburg,Virginia,USA,22-26March2009,http://www.edd.uio.no/artiklar/arkeologi/holmen_ore_caa2009.pdf

HolmenJ.,OreC.&EideO.(2004):DocumentingTwoHistoriesatOnce:DiggingintoArchaeology.Proceedingsofthe30thComputerApplicationsandQuantitativeMethodsinArchaeology,vol.1227ofBARInternationalSeries,Oxford:Archaeopress.

HongY.,SolankiM.,FoxhallL.&QuerciaA.(2010):AFrameworkforTransformingArchaeologicalDatabasestoOntologicalDatasets.In:Proceedingsofthe38thInternationalConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology(CAA).Granada,Spain,April2010,http://www.tracingnetworks.ac.uk/publications/CAA2010/paper.pdf



Horne,Ryan(2014):BeyondMapsasImagesattheAncientWorldMappingCenter.ISAWPaper7.9,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

HoxhaJ.,RulaA.&EllB.(2011):Towardsgreenlinkeddata.COLD2011-2ndInternationalWorkshoponConsumingLinkedData@ISWC2011proceedings,http://ceur-ws.org/Vol-782/HoxhaEtAl_COLD2011.pdf

Huebner,Katherine(2009):Howtaxonomicrevisionsaffecttheinterpretationofspecimenidentificationinbiologicalfielddata.In:mURJ,4(1):25-29,http://msurj.mcgill.ca/vol4/iss1/Huebner2009.pdf

Huggett,Jeremy(2012):PromiseandParadox:AccessingOpenDatainArchaeology.ProceedingsoftheDigitalHumanitiesCongress2012,EditedbyC.Mills,M.Pidd&E.Ward,http://www.hrionline.ac.uk/openbook/chapter/dhc2012-huggett

HuijboomN.&VandenBroekT.(2011):Opendata:aninternationalcomparisonofstrategies.In:EuropeanJournalofePractice,Issue12,March/April2011,pp.4-15,https://joinup.ec.europa.eu/sites/default/files/76/a7/05/ePractice%20Journal-%20Vol.%2012-March_April%202011.pdf

HUMA-NUM-latrèsgrandeinfrastructurederecherchedeshumanitésnumérique,http://www.huma-num.fr

HunterJ.&GerberA.(2010):Harvestingcommunityannotationson3Dmodelsofmuseumartefactstoenhanceknowledge,discoveryandre-use.In:JournalofCulturalHeritage,11(1):81-90,https://www.researchgate.net/search.Search.html?query=Harvesting+community+annotations+on+3D+models+of+museum+artefacts+to+enhance+knowledge%2C+discovery+and+re-use

HunterJ.&GerberA.(2012):TowardsAnnotopia-EnablingtheSemanticInteroperabilityofWeb-BasedAnnotations.FutureInternet,4(3):788-806,http://www.mdpi.com/1999-5903/4/3/788

HunterJ.&YuC.-H.(2011):AssessingtheValueofSemanticAnnotationServicesfor3DMuseumArtefacts.SustainableDatafromDigitalResearchConference(SDDR2011),Melbourne,13-14December2011(authors’manuscript),http://ses.library.usyd.edu.au/bitstream/2123/7951/1/HunterYu.pdf

HunterJ.,KhanI.&GerberA.(2008):HarVANA-HarvestingCommunityTagstoEnrichCollectionMetadata.JointConferenceonDigitalLibraries,JCDL2008.Pittsburgh,USA,16-20June2008,http://www.itee.uq.edu.au/eresearch/filething/files/get/papers/2008/Hunter_JCDL2008.pdf

HylandB.&Villazón-TerrazasB.(eds.,2011):CookbookforOpenGovernmentLinkedData.RevisedVersion,December2011,https://www.w3.org/2011/gld/wiki/Linked_Data_Cookbook

Hyland,Bernadette(2010):PreparingforaLinkedDataEnterprise,in:Wood,David(ed.,2010):LinkingEnterpriseData.Springer,manuscripthttp://linkeddatadeveloper.com/Projects/Linking-Enterprise-Data/Manuscript/led-hyland.html

HyvönenE.,IkkalaE.&TuominenJ.(2016):LinkedDatabrokeringserviceforhistoricalplacesandmaps,pp.39-52,in:ProceedingsofWHiSe2016-1stWorkshoponHumanitiesintheSemanticWeb,Anissaras,Greece,29May2016,http://ceur-ws.org/Vol-1608/paper-06.pdf

HyvönenE.,KettulaS.,RaatikkaV.etal.(2002).SemanticinteroperabilityontheWeb.CaseFinnishMuseumsOn-line.In:HyvönenE.&KlemettinenM.(2002):TowadstheSemanticWebandWebServices.ProceedingsoftheXMLFinland2002Conference,Helsinki,Finland,21-22October2002,http://www.cs.helsinki.fi/u/eahyvone/xmlfinland2002/ProceedingsXML2002-final.pdf



HyvönenE.,LindquistT.,TörnroosJ.&MäkeläE.(2012):Historyonthesemanticwebaslinkeddata:aneventgazetteerandtimelinefortheWorldWarI.ProceedingsofCIDOC2012,EnrichingCulturalHeritage,10–14June2012,Helsinki,Finland,http://www.cidoc2012.fi/en/File/1609/hyvonen.pdf

HyvönenE.,MäkeläE.,KauppinenT.etal.(2009a):CultureSampo-FinnishCulturalHeritageCollectionsontheSemanticWeb2.0.InProceedingsofthe1stInternationalSymposiumonDigitalhumanitiesforJapaneseArtsandCultures,RitsumeikanUniversity,Kyoto,Japan,March2009,http://www.seco.tkk.fi/publications/2009/hyvonen-et-al-culturesampo-dh-jac-2009.pdf

HyvönenE.,MäkeläE.,KauppinenT.etal.(2009b):CultureSampo:ANationalPublicationSystemofCulturalHeritageontheSemanticWeb2.0.6thEuropeanSemanticWebConference,proceedings(LectureNotesinComputerScience,vol.5554/2009):851–856,http://www.seco.tkk.fi/publications/2009/hyvonen-et-al-culsa-demo-eswc-2009.pdf

HyvönenE.,MäkeläE.,KauppinenT.etal.(2009c):FinnishCultureontheSemanticWeb2.0.ThematicPerspectivesfortheEnd-user.In:MuseumsandtheWeb,volume2009,http://www.archimuse.com/mw2009/papers/hyvonen/hyvonen.html

HyvönenE.,MäkeläE.,SalminenM.etal.(2005):MuseumFinland-FinnishMuseumsontheSemanticWeb.JournalofWebSemantics,3(2):224–241,http://www.seco.tkk.fi/publications/2005/hyvonen-makela-et-al-museumfinland-finnish-2005.pdf

HyvönenE.,SaarelaS.&ViljanenK.(2004):ApplicationofOntologyTechniquestoView-BasedSemanticSearchandBrowsing.Proceedingsofthe1stEuropeanSemanticWebSymposium(LectureNotesinComputerScience,vol.2053):92–106,http://www.seco.tkk.fi/publications/2004/hyvonen-saarela-et-al-application-of-ontology-techniques-2004.pdf

HyvönenE.,SaarelaS.,ViljanenK.etal.(2004):ACulturalCommunityPortalforPublishingMuseumCollectionsontheSemanticWeb.ECAIWorkshoponApplicationofSemanticWebTechnologiestoWebCommunities(CEURWorkshopProceedings,vol.107)http://sunsite.informatik.rwth-aachen.de/Publications/CEUR-WS/Vol-107/paper8.pdf

HyvönenE.,TuominenJ.,AlonenM.&MäkeläE.(2014):LinkedDataFinland:A7-starModelandPlatformforPublishingandRe-usingLinkedDatasets.ProceedingsofESWC2014DemoandPosterPapers,Springer-Verlag,http://seco.cs.aalto.fi/publications/2014/hyvonen-et-al-ldf-2014.pdf

HyvönenE.,ViljanenK.,TuominenJ.&SeppäläK.(2008):BuildingaNationalSemanticWebOntologyandOntologyServiceInfrastructure-theFinnONTOApproach,pp.95–109,in:5thEuropeanSemanticWebConferenceproceedings(LectureNotesinComputerScience,vol.5021),http://www.seco.tkk.fi/publications/2008/hyvonen-et-al-building-2008.pdf

Hyvönen,Eero(2009):SemanticPortalsforCulturalHeritage.HandbookonOntologies.2ndedition;chapter,http://www.seco.tkk.fi/publications/2009/hyvonen-portals-2009.pdf

Hyvönen,Eero(2012):PublishingandUsingCulturalHeritageLinkedDataontheSemanticWeb.SynthesisLecturesontheSemanticWeb:TheoryandTechnology.PaloAlto:Morgan&Claypool,http://www.seco.tkk.fi/publications/2012/hyvonen-ch-book-2012.pdf

ICOM(2011):ICOMrecommendationonLinkedOpenDataformuseums,http://network.icom.museum/fileadmin/user_upload/minisites/cidoc/AGM_2011/LoD_For_Museums%20v1.6.pdf



ICONCLASSasLinkedOpenData,http://www.iconclass.org/help/lod

iDAI.gazetteer(GermanArchaeologicalInstitute),http://gazetteer.dainst.org

InstitutefortheStudyoftheAncientWorld(ISAW),http://isaw.nyu.edu

InventaireNationalduPatrimoineNaturel/NationalInventoryofNaturalHeritage(Muséumnationald’Histoirenaturelle),http://inpn.mnhn.fr

InventairesarchéozoologiquesetarchéobotaniquesdeFrance-I2AF(Muséumnationald’Histoirenaturelle),https://inpn.mnhn.fr/espece/inventaire/I100

IrishNationalMonumentsServicemonumentclasslist,http://webgis.archaeology.ie/NationalMonuments/WebServiceQuery/Lookup.aspx

ISA-InteroperabilitySolutionsforEuropeanPublicAdministrations(2013):CookbookfortranslatingrelationaldatamodelstoRDFschemas.PreparedfortheISAprogrammebyPwCEUServices,27/02/2013,http://ec.europa.eu/isa/documents/cookbook-for-rdf-schemas-v2.pdf

ISA-InteroperabilitySolutionsforEuropeanPublicAdministrations(2012):StudyonpersistentURIs,withidentificationofbestpracticesandrecommendationsonthetopicfortheMSsandtheEC.PreparedbyP.Archer(W3C/ERCIM),S.GoedertierandN.Loutas(PwCEUServices).ProjectdeliverableD7.1.3,December2012,https://joinup.ec.europa.eu/community/semic/document/10-rules-persistent-uris

IsaacA.&HaslhoferB.(2013):EuropeanaLinkedOpenData–data.europeana.eu.SemanticWebJournal,4(3):291-297,http://www.semantic-web-journal.net/system/files/swj297_1.pdf

IsaacA.,ClayphanR.&HaslhoferB.(2012):Europeana:MovingtoLinkedOpenData.InformationStandardsQuarterly,2012Spring/Summer,24(2/3):34-40,http://www.niso.org/apps/group_public/download.php/9407/IP_Isaac-etal_Europeana_isqv24no2-3.pdf

IsaacA.,WaitesW.,YoungJ.&ZengM.(eds.,2011):LibraryLinkedDataIncubatorGroup:Datasets,valuevocabularies,andmetadataelementsets[W3CIncubatorGroupReport,October25,2011].http://www.w3.org/2005/Incubator/lld/XGR-lld-vocabdataset/

IsaksenL.,BarkerE.,KansaE.&ByrneK.(2011):GooglingAncientPlaces.ProceedingsofDigitalHumanities2011(DH2011),Stanford,CA,June2011(onlinepaper),http://dh2011abstracts.stanford.edu/xtf/view?docId=tei/ab-349.xml;query=;brand=default

IsaksenL.,MartinezK.&EarlG.(2010b):Interoperatewithwhom?Archaeology,Formality&theSemanticWeb.CAAUKChapterMeeting,UCL,London,19-20February2010,http://leifuss.files.wordpress.com/2008/08/caauk2010isaksen1.pdf

IsaksenL.,MartinezK.&EarlG.(2011):SemanticTechnologiesinCulturalHeritage.Past,PresentandFuture.CulturalHeritage&theSemanticWeb.BritishMuseum,London(slides:http://leifuss.files.wordpress.com/2008/08/bmisaksenfinalslides.pdf)

IsaksenL.,MartinezK.,GibbinsN.&EarlG.&KeayS.(2010a):InteroperateWithWhom?Formality,ArchaeologyandtheSemanticWeb(poster).WebScienceConference2010(WebSci10),Raleigh,USA,26-27April2010,http://eprints.soton.ac.uk/150319/

IsaksenL.,MartinezK.,GibbinsN.,EarlG.&KeayS.(2009):LinkingArchaeologicalData,in:FrischerB.etal.(eds.):MakingHistoryInteractive.CAA2009,Williamsburg,Virginia,Archaeopress,Oxford,pp.130-136,http://proceedings.caaconference.org/files/2009/18_Isaksen_et_al_CAA2009.pdf



IsaksenL.,RainerS.,deSotoCañamaresP.&BarkerE.T(2016):PelagiosCommons:DecentralizingtheWebofHistoricalData.PresentationatCAA2016Oslo,session“LinkedPasts:ConnectingIslandsofContent”,30March2016(paperforthcoming)

IsaksenL.,SimonR.,BarkerE.&deSotoCañamaresP.(2014):Pelagiosandtheemerginggraphofancientworlddata,pp.197-201,in:WebSci'14-Proceedingsofthe2014ACMConferenceonWebScience,IndianaUniversity,Bloomington,23-26June2014;preprint,http://www.researchgate.net/publication/266659779_Pelagios_and_the_emerging_graph_of_ancient_world_data

Isaksen,Leif(2011):ArchaeologyandtheSemanticWeb.Thesis,UniversityofSouthampton,SchoolofElectronicsandComputerScience,December2011,http://eprints.soton.ac.uk/206421/1/Thesis.pdf

Ivanov,Vladimir(2011):TheOpenKunstkammerDataProject.In:ERCIMNews,86,July2011,43-44,http://ercim-news.ercim.eu/images/stories/EN86/EN86-web.pdf(seealso:http://data.kunstkamera.ru)

JankowskiJ.,CobosY.,HausenblasM.&DeckerS.(2009):AccessingCulturalHeritageusingtheWebofData[CHoWDer-CulturalHeritageontheWebofData].In:VAST’09-10thInternationalSymposiumonVirtualReality,ArchaeologyandCulturalHeritage2009,St.Julians,Malta,http://aran.library.nuigalway.ie/xmlui/bitstream/handle/10379/455/VAST2009-CHoWDer.pdf?sequence=1

JanowiczK.,HitzlerP.,AdamsB.,KolasD.&VardemanC.(2014):FiveStarsofLinkedDataVocabularyUse.In:SemanticWebJournal,5(3):173-176;http://www.semantic-web-journal.net/content/five-stars-linked-data-vocabulary-use

JanowiczK.,ScheiderS.,PehleT.&HartG.(2012):GeospatialSemanticsandLinkedSpatiotemporalData-Past,Present,andFuture.In:SemanticWebJournal,3(4):321-332;http://www.semantic-web-journal.net/sites/default/files/swj330_0.pdf

Janowicz,Krzysztof(2009):TheRoleofPlacefortheSpatialReferencingofHeritageData.WorkshoponTheCulturalHeritageofHistoricEuropeanCitiesandPublicParticipatoryGIS.UniversityofYork,September2009,http://geog.ucsb.edu/~jano/chwy09_janowicz.pdf

Jansma,Esther(2013):Towardssustainabilityindendroarchaeology:thepreservation,linkageandreuseoftree-ringdatafromtheculturalandnaturalheritageinEurope,pp.169-176,in:Bleicher,Nielsetal.(eds.):DENDRO-Chronologie-Typologie-Ökologie.Freiburg:Janus(paperavailableonwww.academia.edu)

JarrettJ.,ZambaniniS.,Hüber-MorkR.&FelicettiA.(2011):Coinage,DigitizationandtheWorld-WideWeb:NumismaticsandtheCOINSProject.In:NewTechnologiesinMedievalandRenaissanceStudies3,459–489;preprint,https://www.academia.edu/2147548/Coinage_Digitization_and_the_World-Wide_Web_numismatics_and_the_COINS_Project

JentzschA.,CyganiakR.&BizerC.(2011):StateoftheLODCloud,September2011,http://lod-cloud.net/state/

JohnsonT.&EstlundK.(2014):RecipesforEnhancingDigitalCollectionswithLinkedData.In:Code4LibJournal,Issue23,17January2014,http://journal.code4lib.org/articles/9214

JonesS.,MacSweenA.,JeffreyS.,MorrisR.&HeyworthM.(2001):Fromthegroundup:Thepublicationofarchaeologicalprojects:auserneedssurvey,http://www.britarch.ac.uk/pubs/puns



JordalE.,UlebergE.&HaugeB.(2012):WasItWorthIt?ExperienceswithaCIDOCCRM-basedDatabase,pp.255-260,in:CAA2011-Proceedingsofthe39thAnnualConferenceofComputerApplicationsandQuantitativeMethodsinArchaeology,Beijing,China,12-16April2011,http://proceedings.caaconference.org/files/2011/28_Jordal_et_al_CAA2011.pdf

JSON-JavaScriptObjectNotation,http://json.org

JSON-LD-JSONforLinkingData,http://json-ld.org

Kamps,Jaap(2015):WhenSearchbecomesResearchandResearchbecomesSearch.KeynotepresentationatSIGIR’13-WorkshoponExploration,NavigationandRetrievalofInformationinCulturalHeritage(ENRICH)1August2013,Dublin,Ireland,http://de.slideshare.net/jaap.kamps/sigir-workshop-enrich13

Kamura,Tetsuroetal.(2011):BuildingLinkedDataforCulturalInformationResourcesinJapan.ProceedingsofMuseumsandtheWeb2011,http://www.museumsandtheweb.com/mw2011/papers/building_linked_data_for_cultural_information_.html

KansaE.&BissellA.(2010):Websyndicationapproachesforsharingprimarydatain“smallscience”domains.In:DataScienceJournal,Volume9:42-53,https://www.jstage.jst.go.jp/article/dsj/9/0/9_009-012/_pdf

KansaE.&Whitcher-KansaS.(2011):EnhancingHumanitiesResearchProductivityinaCollaborativeDataSharingEnvironment.WhitePapertotheNEHDivisionofPreservationandAccess,27June2011,http://alexandriaarchive.org/wp-content/uploads/2011/09/white_paper_PK_50072.pdf

KansaE.&Whitcher-KansaS.(2013):Weallknowthata14isasheep:datapublicationandprofessionalisminarchaeologicalcommunication.In:JournalofEasternMediterraneanArchaeologyandHeritageStudies,1(1):88–97;preprint,https://escholarship.org/uc/item/9m48q1ff

KansaE.,Whitcher-KansaS.&ArbuckleB.(2014):PublishingandPushing:MixingModelsforCommunicatingResearchDatainArchaeology.In:InternationalJournalforDigitalCuration,9(1),http://www.ijdc.net/index.php/ijdc/article/view/9.1.57/341

KansaE.,Whitcher-KansaS.&WatrallE.(eds.,2011):Archaeology2.0:NewApproachestoCommunicationandCollaboration.CotsenInstituteofArchaeology,UCLosAngeles,http://escholarship.org/uc/item/1r6137tb

Kansa,Eric(2014a):OpenContextandLinkedData.ISAWPaper7.10,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/kansa/

Kansa,Eric(2014b):LinkedData,Publication,andtheLifeCycleofArchaeologicalInformation.Presentationat8.DeutscherArchäologiekongress,Berlin,8October2014,http://www.ianus-fdz.de/attachments/download/697//06_Kansa_OpenContext.pdf

Kansa,Eric(2015):ContextualizingDigitalDataasScholarshipinEasternMediterraneanArchaeology.In:CHSResearchBulletin,3(2),http://nrs.harvard.edu/urn-3:hlnc.essay:KansaE.Contextualizing_Digital_Data_as_Scholarship.2015

Katz,DanielS.etal.(2014):SummaryoftheFirstWorkshoponSustainableSoftwareforScience:PracticeandExperiences(WSSSPE1).In:JournalofOpenResearchSoftware,2(1):e6:1-21,http://dx.doi.org/10.5334/jors.an

KauppinenT.,BaglatziA.&KeßlerC.(2013):LinkedScience:InterconnectingScientificAssets.In:CritchlowT.&Kleese-VanDamK.(eds.):DataIntensiveScience.CRCPress;preprint,



http://linkedscience.org/wp-content/uploads/2012/02/linked-science-bookchapter-revised-2011-11-16.pdf

Kerameikos,http://kerameikos.org

Kintigh,Keith(2006):Thechallengeofarchaeologicaldataintegration.PaperpresentedatthemeetingoftheUnionInternationaledesSciencesPréhistoriquesetProtohistoriques,sessionTechnologyandMethodologyforArchaeologicalPractice,Lisbon,September2006,http://archaeoinformatics.org/articles/Kintigh2006UISPP.pdf

KiryakovA.,OgnyanoffD.,VelkovR.,TashevZ.&PeikovI.(2009):LDSR:MaterializedReason-ableViewtotheWebofLinkedData.In:Proceedingsofthe3rdInternationalRuleML-2009Challenge.LasVegas,USA,http://ceur-ws.org/Vol-549/paper9.pdf

KobilarovG.,ScottT.,RaimondY.etal.(2009):MediaMeetsSemanticWeb–HowtheBBCUsesDBpediaandLinkedDatatoMakeConnections.In:L.Aroyoetal.(Eds.):ESWC2009,LNCS5554,BerlinandHeidelberg:Springer2009,pp.723–737,http://derivadow.files.wordpress.com/2009/06/eswc2009-bbc-dbpedia-2.pdf

KondertF.,SchandlT.&BlumauerA.(2011):Docontrolledvocabulariesmatter?Surveyresults.SemanticWebCompany,Vienna,June2011,http://www.semantic-web.at/sites/default/files/files/Survey_Do_Controlled_Vocabularies_Matter_2011_June_0.pdf

KosemI.,JakubičekM.,KallasJ.&KrekS.(eds.,2015):ElectronicLexicographyinthe21stCentury:LinkingLexicalDataintheDigitalAge.ProceedingsofeLex2015-ElectronicLexicographyinthe21stcentury:LinkingLexicalData,HerstmonceuxCastle,Sussex,UK,11-13August2015,https://elex.link/elex2015/conference-proceedings/

Krueger,KristiJ.(2013):ACaseStudyofAssertionsfortheIronAgeandImplicationsforTemporalMetadataCreation.AMaster’sPaperfortheM.S.inL.S.degree.UniversityofNorthCarolinaatChapelHill,April2013,https://cdr.lib.unc.edu/record/uuid:a8f56c09-954c-45ca-931b-a7fc2bf51dd5

Lana,Maurizio(2014):Geolat:GeographyforLatinLiterature.ISAWPaper7.11,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

LangM.,CarverF.&PrintzS.(2013):StandardisedVocabularyinArchaeologicalDatabases,pp.468-473,in:CAA2012Southampton,VolumeII,AmsterdamUniversityPress,http://dare.uva.nl/cgi/arno/show.cgi?fid=545855

Lange,A.G.(ed.,2004):ReferenceCollections.FoundationforFutureArchaeology.ProceedingsoftheinternationalconferenceontheEuropeanelectronicReferenceCollection,12-13May2004,ROB,Amersfoort,TheNetherlands,http://cultureelerfgoed.nl/sites/default/files/publications/reference-collections-foundation_for_future_archaeology.pdf

LATC-LODAroundTheClock(2012):FinalReleaseofP&CLibrary.ProjectdeliverableD3.3.1,29February2012,http://latc-project.eu/node/89

LAWD-LinkingAncientWorldDataontology,https://github.com/lawdi/LAWD

LAWDI-LinkedAncientWorldDataInstitute(USA,NEH-fundedproject,2012-2013),http://wiki.digitalclassicist.org/Linked_Ancient_World_Data_Institute

LeCornecRocheloisC.&IssacF.(2015):WhatTermstoExpresstheCategoriesofNaturalSciencesintheDictionaryofMedievalScientificFrench?,pp.29-42,in:SWASH2016-1stWorkshopon



SemanticWebforScientificHeritage,Portoroz,Slovenia,1June2015,http://ceur-ws.org/Vol-1364/sw4sh-2015.pdf

LeGoffE.,MarletO.,RodierX.,CuretS.&HusiP.(2015):TheinteroperabilityoftheArSol(ArchivesduSol)database:BasedontheCIDOC-CRMontology,pp.179-186,in:CAA2014Paris-Proceedingsofthe42ndAnnualConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology,Paris,France,22-25April2014,Archaeopress,http://www.archaeopress.com/ArchaeopressShop/Public/download.asp?id={5CACE285-4C48-41AE-809E-E98B65C9E4CD}

LedlA.&VoßJ.(2016):DescribingKnowledgeOrganizationSystemsinBARTOCandJSKOS.In:12thInternationalConferenceonTerminologyandKnowledgeEngineering(TKE2016),Copenhagen,22-24June2016;preprint,http://eprints.rclis.org/29366/1/Ledl_Voss_TKE2016_final_version_20160518.pdf

lemon-TheLexiconModelforOntologies(seealso:OntoLexmodel,Cimianoetal.2016),http://lemon-model.net

LiAM-LinkedArchivalMetadataproject(USA,10/2012-9/2013,ledbyTuftsUniversity,DigitalCollectionsandArchives),http://sites.tufts.edu/liam/

LibraryLinkedDataIncubatorGroup(2011):Datasets,ValueVocabularies,andMetadataElementSets.W3CIncubatorGroupReport,25October2011,http://www.w3.org/2005/Incubator/lld/XGR-lld-vocabdataset-20111025/

LibraryofCongress:LinkedDataService,http://id.loc.gov

LIDER-LingHub-LinguisticLinkedOpenDatacloud,http://linghub.lider-project.eu/llod-cloud

LIDER-LingHub(languageresources),http://linghub.lider-project.eu

LIDER-LinkedDataasanenablerofcross-mediaandmultilingualcontentanalyticsforenterprisesacrossEurope(EU,FP7,11/2013-12/2015,http://www.lider-project.eu

Limp,FredrickW.(2011):Web2.0andBeyond,orOntheWeb,NobodyKnowsYou’reanArchaeologist,pp.265-280,in:KansaE.etal.(eds.):Archaeology2.0:NewApproachestoCommunicationandCollaboration.CotsenInstituteofArchaeology,UCLosAngeles,http://escholarship.org/uc/item/1r6137tb

Lincoln,MatthewD.(2016):LinkedOpenRealities:TheJoysandPainsofUsingLODforResearch.In:ArtHistoryandDigitalResearchweblog,6June2016,http://matthewlincoln.net/2016/06/06/linked-open-realities-the-joys-and-pains-of-using-lod-for-research.html

LinkedAncientWorldData:RelatingthePast(2016).PanelatDigitalHumanities2016conference,Kraków,Poland,11-16July2016,http://dh2016.adho.org/abstracts/262

LinkedHeritage&Athena(2011):YourterminologyaspartoftheSemanticWeb.Recommendationsfordesignandmanagement.November2011,http://www.linkedheritage.eu/getFile.php?id=244

LinkedHeritage(EU,ICT-PSP,2011-2013),http://www.linkedheritage.eu

LinkedOpenVocabularies–LOV(OpenKnowledgeFoundation),http://lov.okfn.org

linkedarc.net,http://linkedarc.net;datasets,https://datahub.io/dataset/linkedarc

LinkedBrainz-MusicBrainzinRDFandSPARQL,http://linkedbrainz.org



LinkedDataTools-Freetools,informationandresourcesforthesemanticweb,http://www.linkeddatatools.com

LinkingOpenDataclouddiagram,http://lod-cloud.net

Liuzzo,Pietro(2016):MappingEpigraphicDatabasestoEpiDoc,pp.149-162,in:OrlandiS.etal.(eds.):DigitalandTraditionalEpigraphyinContext.ProceedingsoftheSecondEAGLEInternationalConference.Rome,27-29January2016,http://www.eagle-network.eu/wp-content/uploads/2016/04/EAGLE%20D2.6_EAGLE%20Second%20International%20Conference%20Proceedings.pdf

Liuzzo,PietroM.(2014):TheEuropeanaNetworkofAncientGreekandLatinEpigraphy(EAGLE).ISAWPaper7.12,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

LOCAH-LinkedArchivesandLinkingLivesprojects(UK,JISC-funded,2010-2012,MimasandUKOL),http://locah.archiveshub.ac.uk

LODBrowserSwitch(offersasetofbrowsers),http://browse.semanticweb.org

LOD2-CreatingKnowledgeoutofInterlinkedData(2011):StateoftheArtAnalysis.Projectdeliverable1.2,16January2011,http://static.lod2.eu/Deliverables/deliverable-1.2.pdf

LOD2-CreatingKnowledgeoutofInterlinkedData(EU,FP7-ICT,2010–2014),http://lod2.eu

LOD-LAM,theInternationalLODinLibraries,Archives,andMuseumsSummit,http://lodlam.net

LODStats(AgileKnowledgeEngineeringandSemanticWebGroupatUniversityofLeipzig,Germany),http://stats.lod2.eu

MacKay,Camilla(2014):BrynMawrClassicalReview.ISAWPaper7.13,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

Madsen,Torsten(2004):Classificationandarchaeologicalknowledgebases,pp.35-42,in:Lange,A.G.(ed.):ReferenceCollections.FoundationforFutureArchaeology.Amersfoort,TheNetherlands:ROB,http://cultureelerfgoed.nl/sites/default/files/publications/reference-collections-foundation_for_future_archaeology.pdf

Mantegari,Glauco(2009):Culturalheritageonthesemanticweb:Fromrepresentationtofruition.Ph.D.dissertation,UniversitàdegliStudidiMilano-Bicocca,QUASIProject,http://boa.unimib.it/bitstream/10281/9184/3/phd_unimib_708063.pdf

MappingMemoryManager-3M(facilitatesthemappingofdatabasestotheextendedCIDOCCRM),FoundationforResearchandTechnologyHellas,InstituteofComputerScience,http://www.ics.forth.gr/isl/3M

MarletO.,CuretS.,RodierX.&Bouchou-MarkhoffB.(2016):UsingCIDOCCRMfordynamicallyqueryingArSol,arelationaldatabase,fromthesemanticweb,pp.241-249,in:CAA2015-KeeptheRevolutionGoing:Proceedingsofthe43rdAnnualConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology.Oxford:Archaeopress,http://archaeopress.com/ArchaeopressShop/Public/download.asp?id={77DEDD4E-DE8F-43A4-B115-ABE0BB038DA7}

MASA-MémoiredesArchéologuesetdesSitesArchéologiques,http://masa.hypotheses.org

MaturanaR.A.,OrtegaM.&López-SolaS.(2013):Mismuseos.net:ArtAfterTechnology.PuttingculturaldatatoworkinaLinkedDataplatform.In:ProceedingsofVeni2013-LinkedUpVeniCompetitiononLinkedandOpenDataforEducation,Geneva,17September2013,http://ceur-ws.org/Vol-1124/linkedup_veni2013_03.pdf



MayK.,BindingC.&TudhopeD.(2015):BarriersandopportunitiesforLinkedOpenDatauseinarchaeologyandculturalheritage.In:ArchäologischeInformationen,Volume38,http://journals.ub.uni-heidelberg.de/index.php/arch-inf/article/view/26162/19880

MayK.,BindingC.&Tudhope,D.(2010):FollowingaSTAR?Sheddingmorelightonsemantictechnologiesforarchaeologicalresources.ComputerApplicationsandQuantitativeMethodsinArchaeology2009(BARIntSer2079),227-233,http://proceedings.caaconference.org/files/2009/28_May_et_al_CAA2009.pdf

MayK.,BindingC.,TudhopeD.&JeffreyS.(2011):SemanticTechnologiesEnhancingLinksandLinkedDataforArchaeologicalResources,pp.261-272,in:CAA2011-RevivethePast.Proceedingsofthe39thAnnualConferenceofComputerApplicationsandQuantitativeMethodsinArchaeology(CAA),Beijing,China,12-16April2011,http://proceedings.caaconference.org/files/2011/29_May_et_al_CAA2011.pdf

May,Keith(2016):TheMatrix:ConnectingTimeandSpacewitharchaeologicalresearchquestionsinvolvingspatio-temporalphenomenaandtheconceptualrelationshipsbetweenthem.PresentationatCAA2016Oslo,session“LinkedPasts:ConnectingIslandsofContent”,30March2016,http://de.slideshare.net/Keith.May/caa-2016-the-matrix-connecting-time-space

MazziniS.&RicciF.(2011):EAC-CPFOntologyandlinkedarchivaldata,pp.72–81,in:Proceedingsofthe1stInternationalWorkshoponSemanticDigitalArchives,29Sept2011,Berlin,Germany.CEURWorkshopProceedings,vol.801,http://ceur-ws.org/Vol-801/paper6.pdf

McCraeJ.P.&CimianoP.(2015):Linghub:aLinkedDatabasedportalsupportingthediscoveryoflanguageresources,pp.88-91,in:SEMANTiCS2015-11thInternationalConferenceonSemanticSystems,ProceedingsofthePostersandDemosTrack,Vienna,Austria,15-17September2015,http://ceur-ws.org/Vol-1481/paper27.pdf

McMichael,A.L.(2014):ByzantineCappadocia:SmallDataandtheDissertation.ISAWPaper7.14,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

MeadowsA.&GruberE.(2014):CoinageandNumismaticMethods.ACaseStudyofLinkingaDiscipline.ISAWPaper7.15,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

Meadows,Andrew(2015):OnlineCoinsoftheRomanEmpire:AnOpenResourceforRomanNumismatics,December2015,https://t.co/pKksMjf7qb

MeeksE.&GrossnerK.(2012):ORBIS:AnInteractiveScholarlyWorkontheRomanWorld.In:JournalofDigitalHumanities,1(3),http://journalofdigitalhumanities.org/1-3/orbis-an-interactive-scholarly-work-on-the-roman-world-by-elijah-meeks-and-karl-grossner/

Meroño-PeñuelaA.,AshkpourA.,RietveldL.,HoekstraR.&SchlobachS.(2012):LinkedHumanitiesData:Thenextfrontier?[censusdata].In:ProceedingsofLISC2012-2ndInternationalWorkshoponLinkedScience2012,Boston,12November2012,http://ceur-ws.org/Vol-951/paper3.pdf

Meroño-PeñuelaA.,AshkpourA.,vanErpM.etal.(2014):SemanticTechnologiesforHistoricalResearch:ASurvey.In:SemanticWebJournal,paper588,http://www.semantic-web-journal.net/system/files/swj588_0.pdf

Meyers,Katy(2014):ExploringanOpportunitytoLinktheDeadinAncientRome.ISAWPaper7.16,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

MichelF.,MontagnatJ.&Faron-ZuckerC.(2013):AsurveyofRDBtoRDFtranslationapproachesandtools.EquipesModalis/Wimmics.RapportdeRecherche,ISRNI3S/RR,2013-04-FR,Novembre



2013,https://hal.inria.fr/file/index/docid/903568/filename/Michel_Montagnat_Faron_2013_-_A_survey_of_RDB_to_RDF_translation_approaches_and_tools.pdf

Miller,Paul(2010):LinkedDataHorizonScan.ReportcommissionedbyJISC.January2010,http://cloudofdata.com/2010/02/final-version-of-linked-data-horizon-scan-now-available-online/

MinadakisN.,MarketakisY.,KondylakisH.,FlourisG.,TheodoridouM.,DoerrM.&deJongG.(2016):X3MLFramework:aneffectivesuiteforsupportingdatamappings,pp.1-12,in:Ronzino,Paola(ed.):Extending,MappingandFocusingtheCRM.ProceedingsoftheEMF-CRMworkshop,Poznan,Poland,17September2015,http://ceur-ws.org/Vol-1656/paper1.pdf

MisMuseos.net:DataHubinformation,http://datahub.io/dataset/mismuseos-gnoss

Missikoff,Oleg(2004):OntologiesasaReferenceFrameworkfortheManagementofKnowledgeintheArchaeologicalDomain,pp.35-39,in:EnterthePast.TheE-WayIntotheFourDimensionsofCulturalHeritage.ArcheoPress;preprint,https://publikationen.uni-tuebingen.de/xmlui/bitstream/handle/10900/60734/02_Missikoff_CAA_2003.pdf?sequence=2&isAllowed=y

Mitchell,ErikT.(2016):TheCurrentStateofLinkedDatainLibraries,Archives,andMuseums.In:ALATechSource-LibraryTechnologyReports,52(1),chapter1,https://journals.ala.org/ltr/article/view/5892/7446

MONDIS-MonumentDamageInformationSystemproject(CzechRepublic),http://www.mondis.cz

MoRe-Metadata&ObjectRepositoryaggregator(ATHENA,DigitalCurationUnit,Greece),http://more.dcu.gr

MorganE.L.etal.(2014):LinkedArchivalMetadata:AGuidebook.Version0.99,23April2014,http://sites.tufts.edu/liam/2014/04/24/version-099/

Morgan,EricL.(2014):LinkedArchivalMetadata:Trendsandgapsinlinkeddataforarchives.LiAM:LinkedArchivalMetadata,http://sites.tufts.edu/liam/2014/04/23/trends/

MouromtsevD.,HaaseP.,ChernyE.,PavlovD.,AndreevA.&SpiridonovaA.(2015):TowardstheRussianLinkedCultureCloud:DataEnrichmentandPublishing,pp.637-651,in:TheSemanticWeb.LatestAdvancesandNewDomains.Springer(LNCP9088);preprint,http://metaphacts.com/images/Papers/Towards-the-Russian-Linked-Culture-Cloud.pdf

MULTITA-CoudyzerE.&LheureuxB.(2015):Multilingualterminologicalresearch(French,DutchandEnglish)forthedevelopmentandintegrationofsemanticallyenrichedscientificthesauri(MULTITA).Summaryoftheresearchproject,30January2015,http://www.belspo.be/belspo/organisation/Publ/pub_ostc/agora/ragLL169sum_en.pdf

MULTITA-Multilingualterminologicalresearch(French,DutchandEnglish)forthedevelopmentandintegrationofsemanticallyenrichedscientificthesauri(7/2012-12/2014),http://www.belspo.be/belspo/fedra/proj.asp?l=fr&COD=AG/LL/169

MungallC.J.,TorniaiC.,GkoutosG.V.,LewisS.E.&HaendelM.A.(2012):Uberon,anintegrativemulti-speciesanatomyontology.GenomeBiology13,R5,http://genomebiology.com/2012/13/1/R5

Murray,William(2014):RAM3DWebPortal.ISAWPaper7.17,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

MuseiItaliani,http://www.linkedopendata.it/datasets/musei



MuseumsandtheMachine-processableWebwiki,editedbyMiaRidge,http://museum-api.pbworks.com/w/page/21933420/Museum%C2%A0APIs

NationalMuseumofIreland:Artefacts,http://www.museum.ie/en/list/artefacts.aspx

NaturalEuropeproject(EU,ICT-PSP,10/2010-09/2013),http://www.natural-europe.eu

NCBIOrganismalClassification,https://bioportal.bioontology.org/ontologies/NCBITAXON

NgongaNgomoA.-C.,AuerS.,LehmannJ.&ZaveriA.(2014):IntroductiontoLinkedDataandItsLifecycleontheWeb,pp.1-99,in:ReasoningWeb.ReasoningontheWebintheBigDataEra.Proceedingsofthe10thInternationalSummerSchool2014,Athens,Greece,8-13September2014.Springer(LNCS8714);preprint,http://jens-lehmann.org/files/2014/reasoning_web_update_linked_data.pdf

NiccolucciF.&HermonS.(2015):Time,chronologyandclassification,pp.265-279,in:BarceloJ.A.&BogdanovicI.(eds.):MathematicsandArchaeology.CRCPress

NiccolucciF.&HermonS.(2016):Representinggazetteersandperiodthesauriinfour-dimensionalspace–time.In:InternationalJournalonDigitalLibraries,17(1):63-69,http://link.springer.com/article/10.1007/s00799-015-0159-x

NiccolucciF.,HermonS.&DoerrM.(2015):Theformallogicalfoundationsofarchaeologicalontologies,pp.86-99,in:BarceloJ.A.&BogdanovicI.(eds.):MathematicsandArchaeology.CRCPress

NikolovA.&d’AquinM.(2011):IdentifyingRelevantSourcesforDataLinkingusingaSemanticWebIndex.LDOW2011,Hyderabad,India,29March2011,http://ceur-ws.org/Vol-813/ldow2011-paper10.pdf

NikolovA.,d’AquinM.&MottaE.(2012):WhatshouldIlinkto?Identifyingrelevantsourcesandclassesfordatalinking,pp.284-299,in:JIST2011-JointInternationalSemanticTechnologyConference.TheSemanticWeb.Springer(LNCS7185);preprint,http://people.kmi.open.ac.uk/andriy/jist2011.pdf

NKOSTaskGroupoftheDublinCoreMetadataInitiative(2015):KOSTypesVocabulary,2015-10-02,http://wiki.dublincore.org/index.php/NKOS_Vocabularies

Nomismaontologyandnumismaticsdatasets,http://nomisma.org

NouvelB.&SinigagliaE.(2014):PACTOLS,unthésauruspourdécrirelesressourcesdocumentairesenarchéologie.MASAConsortium,weblog,17November2014,http://masa.hypotheses.org/116;slides:https://f.hypotheses.org/wp-content/blogs.dir/1718/files/2014/11/01_PACTOLS_MASA20141013.pdf

Nouvel,Blandine(2015):Desoutilsd’enrichissementdocumentairemultilinguespourl’archéologie.MASAweblog,14December2015,http://masa.hypotheses.org/date/2015/12

NowakK.&BonB.(2015):medialatinitas.eu.TowardsShallowIntegrationofLexical,TextualandEncyclopaedicResourcesforLatin,pp.152-169,in:ProceedingsofeLex2015-ElectronicLexicographyinthe21stcentury:LinkingLexicalData,HerstmonceuxCastle,Sussex,UK,11-13August2015,https://elex.link/elex2015/proceedings/eLex_2015_10_Nowak+Bon.pdf

Nurmikko-Fuller,Terhi(2014):AssessingtheSuitabilityofExistingOWLOntologiesfortheRepresentationofNarrativeStructuresinSumerianLiterature.ISAWPaper7.18,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/



NußbaumerP.&HaslhoferB.(2007):CIDOCCRMinAction–ExperiencesandChallenges.Posteratthe11thEuropeanConferenceonResearchandAdvancedTechnologyforDigitalLibraries(ECDL07),Budapest,http://eprints.cs.univie.ac.at/403/1/cidoc_crm_poster_ecdl2007.pdf

NußbaumerP.,HaslhoferB.&KlasW.(2010):TowardsModelImplementationGuidelinesfortheCIDOCConceptualReferenceModel.TechnicalReportTR-201.UniversityofVienna,http://eprints.cs.univie.ac.at/58/

OCLC-OnlineComputerLibraryCenter:LinkedData,http://oclc.org/developer/develop/linked-data.en.html

OldmanD.&RahtzS.(2014):AligningtheAcademywiththeCulturalHeritageSectorthroughtheCIDOCCRMandSemanticWebtechnology,p.80,in:CAA2014Paris,Bookofabstracts,http://caa2014.sciencesconf.org/conference/caa2014/pages/BOACAA_2016.pdf

OldmanD.,DoerrM.&GradmannS.(2015):ZENandtheArtofLinkedData.NewStrategiesforaSemanticWebofHumanistKnowledge,Chapter18inSchreibmanS.,SiemensR.&UnsworthJ.(eds.):ANewCompaniontoDigitalHumanities.Blackwell;preprint,https://www.academia.edu/12608990/ZEN_and_the_Art_of_Linked_Data_New_Strategies_for_a_Semantic_Web_of_Humanist_Knowledge

OldmanD.,DoerrM.,deJongG.,NortonB.&WikmanT.(2014):RealizingLessonsoftheLast20Years:AManifestoforDataProvisioning&AggregationServicesfortheDigitalHumanities(APositionPaper).In:D-LibMagazine,20(7/8),http://www.dlib.org/dlib/july14/oldman/07oldman.html

Oldman,Dominic(2012):TheBritishMuseum,CIDOCCRMandtheShapingofKnowledge.DominicOldmanweblog,4September2012,http://www.oldman.me.uk/blog/the-british-museum-cidoc-crm-and-the-shaping-of-knowledge

Olsson,CarlA.(2016):ALinked(Open)DatahubattheNorwegianDirectorateforCulturalHeritage–acasestudy.PresentationatCAA2016Oslo,session“LinkedPasts:ConnectingIslandsofContent”,30March2016(paperforthcoming)

Omelayenko,Borys(2008):PortingCulturalRepositoriestotheSemanticWeb,pp.14-35,in:KolliasS.&CousinsJ.(eds.):SemanticInteroperabilityintheEuropeanDigitalLibrary.ProceedingsoftheFirstInternationalWorkshop,SIEDL2008,Tenerife,2June2008,http://image.ntua.gr/swamm2006/SIEDLproceedings.pdf

ONKI-FinnishOntologyLibraryService,http://onki.fi

OnlineCoinsoftheRomanEmpire(OCRE),http://numismatics.org/ocre/

ONTOCOM-OntologyCostEstimationwithONTOCOM,http://ontocom.sti-innsbruck.at

Ontop,platformtoquerydatabasesasVirtualRDFGraphsusingSPARQL(UniversityofBozen-Bolzano,KRDBresearchgroup),http://ontop.inf.unibz.it

OomenJ.,BaltussenL.-B.&VanErpM.(2012):Sharingculturalheritagethelinkedopendataway:whyyoushouldsignup.In:MuseumsandtheWeb2012,SanDiego,11-14April2012,http://www.museumsandtheweb.com/mw2012/papers/sharing_cultural_heritage_the_linked_open_data

OpenAnnotationCollaboration,http://www.openannotation.org

OpenArchivesInitiative-ProtocolforMetadataHarvesting(OAI-PMH),http://www.openarchives.org/pmh/



OpenContext:Linkeddataprojects,http://alexandriaarchive.org/projects/linked-data/

OpenDataBarometer(internationalsurveyofopengovernmentaldata),http://opendatabarometer.org

OpenDataCommons(ODC)licenses,http://opendatacommons.org/licenses/

OpenRefine,http://openrefine.org

ORBIS-TheStanfordGeospatialNetworkModeloftheRomanWorld,http://orbis.stanford.edu

OrdnanceSurvey(UK),http://data.ordnancesurvey.co.uk

OrlandiS.,SantucciR.,CasarosaV.&LiuzzoP.M.(2014):InformationTechnologiesforEpigraphyandCulturalHeritage.ProceedingsoftheFirstEAGLEInternationalConference,Paris,http://www.eagle-network.eu/wp-content/uploads/2015/01/Paris-Conference-Proceedings.pdf

PACTOLS-Peuples,Anthroponymes,Chronologie,Toponymes,Oeuvres,LieuxetSujets(thesaurus),http://frantiq.mom.fr/thesaurus-pactols

Page,Roderic(2009):SemanticPublishing:towardsrealintegrationbylinking.iPhyloweblog,20April2009,http://iphylo.blogspot.co.at/2009/04/semantic-publishing-towards-real.html

PanX.,SchifferT.,HecherM.etal.(2012a):AscalablerepositoryinfrastructureforCHdigitalobjectmanagement.In:18thInternationalConferenceonVirtualSystemsandMultimedia,Milan,Italy,September2012,http://havemann.cgv.tugraz.at/Publications/2012_PSHx12__ScalableRepositoryInfrastructureForCHObjectManagement.pdf

PanX.,SchifferT.,SchröttnerM.etal.(2012b):Anenhanceddistributedrepositoryforworkingwith3dassetsinculturalheritage.In:4thInternationalEuro-MediterraneanConferenceonDigitalHeritage(EuroMed),Limassol,Cyprus,October2012.SpringerLNCS,http://link.springer.com/chapter/10.1007%2F978-3-642-34234-9_35

ParryR.,PooleN.&PrattyJ.(2008):SemanticDissonance:DoWeNeed(andDoWeUnderstand)theSemanticWeb?ProceedingsofMuseumsandtheWebConference2008,http://www.archimuse.com/mw2008/papers/parry/parry.html

PATHS-PersonalisedAccesstoCulturalHeritageSpaces(EU,FP7project,01/2011-12/2013),http://www.paths-project.eu

PatroumpasK.,AlexakisM.,GiannopoulosG.&AthanasiouS.(2014):TripleGeo:anETLToolforTransformingGeospatialDataintoRDFTriples,pp.275-278,in:ProceedingsoftheWorkshopsoftheEDBT/ICDT2014JointConference,Athens,Greece,28March2014,http://ceur-ws.org/Vol-1133/paper-44.pdf

PearceL.&SchmitzP.(2014):BerkeleyProsopographyServices.ISAWPaper7.19,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

Pelagiosproject,http://commons.pelagios.org

Pelagios:JoiningPelagios,https://github.com/pelagios/pelagios-cookbook/wiki/Joining-Pelagios

PenaSernaS.,SchmedtH.,RitzM.&StorkA.(2012):InteractiveSemanticEnrichmentof3DCulturalHeritageCollections.In:VAST’12-The13thInternationalSymposiumonVirtualReality,ArchaeologyandCulturalHeritage,Brighton,UK,http://culturalinformatics.org.uk/files/papers/InteractiveSemanticEnrichment2012.pdf



PenaSernaS.,ScopignoR.,DoerrM.etal.(2011):3D-centredmedialinkingandsemanticenrichmentthroughintegratedsearching,browsing,viewingandannotating.VAST11:12thInternationalSymposiumonVirtualReality,ArchaeologyandIntelligentCulturalHeritage,Prato,Italy(notopenlyavailableonline)

PeriodO-Periods,Organizedproject,http://perio.do

Pett,Daniel(2014a):LinkingPortableAntiquitiestoawiderweb.ISAWPaper7.20,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/pett/

Pett,Daniel(2014b):MakingthelinkstoPortableAntiquitiesSchemedata.In:CAA2014Paris,Bookofabstracts,p.81,http://f.hypotheses.org/wp-content/blogs.dir/1309/files/2014/04/CAA2014-BOA-S07-20140424.pdf

Pett,Daniel(n.d.):ImplementingLinkedDatawithinthePortableAntiquitiesScheme,https://www.academia.edu/9347715/Implementing_Linked_Data_within_the_Portable_Antiquities_Scheme

PICOthesaurus(CentralInstitutefortheUnionCatalogue-ICCU,Italy,http://purl.org/pico/thesaurus_4.2.0.skos.xml

PittiD.V.,PopoviciB.F.,StocktingW.&ClavaudF.(2014):ExpertsGrouponArchivalDescription:InterimReport.Girona2014:ArxiusIIndustriesCulturals.Girona2014:ArxiusiIndústriesCulturals,Girona,Spain,11-15October2014,http://www.girona.cat/web/ica2014/ponents/textos/id56.pdf

PlacenamesDatabaseofIreland,http://www.logainm.ie/en/

PlanetData(2012):Conceptualmodelandbestpracticesforhigh-qualitymetadatapublishing.ProjectdeliverableD2.1,http://planet-data-wiki.sti2.at/web/File:D2.1.pdf

Pleiades-GazetteeroftheAncientWorld,http://pleiades.stoa.org

Poehler,Eric(2014):PompeiiBibliographyandMappingResource.ISAWPaper7.21,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

PortableAntiquitiesScheme,http://finds.org.uk

Portnoy,David(2014):WhatHappenedtotheSemanticWeb?September2014,http://david.portnoy.us/what-happened-to-the-semantic-web/

PricewaterhouseCoopers(2009):TechnologyForecast.Spring2009,http://www.pwc.com/us/en/technology-forecast/spring2009/

Rabinowitz,Adam(2014):It’sabouttime:HistoricalPeriodizationandLinkedAncientWorldData.ISAWPaper7.22,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/rabinowitz/

RaimondY.,SmethurstM.,McParlandA.&LowisC.(2013):UsingthePasttoExplainthePresent:InterlinkingCurrentAffairswithArchivesviatheSemanticWeb,pp.146-161,in:TheSemanticWeb–ISWC2013,12thInternationalSemanticWebConference,Sydney,21-25October2013,PartII,Springer(LNCS8219);preprint,http://downloads.bbc.co.uk/rd/pubs/whp/whp-pdf-files/WHP260.pdf

RakhmawatiN.A.,UmbrichJ.,KarnstedtM.,HasnainA.&HausenblasM.(2013):QueryingoverFederatedSPARQLEndpoints|AStateoftheArtSurvey.DERITechnicalReport2013-06-07,June2013,http://www.deri.ie/sites/default/files/publications/1306.1723v1.pdf



Reinhard,Andrew(2014):PublishingArchaeologicalLinkedOpenData:FromSteampunktoSustainability.ISAWPaper7.23,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

ReLoad-RepositoryforLinkedOpenArchivalData(Italy,2010-2013,ArchivioCentraledelloStato,IstitutoperiBeniculturalidell’Emilia-Romagnaandregesta.exe),http://labs.regesta.com/progettoReload/

ReLoad(2013):ProjectdescriptionforLODLAM2013summit,http://summit2013.lodlam.net/2012/12/01/challenge-entry-reload-repository-for-linked-open-archival-data/

ResearchSpace-CreatingtheCulturalHeritageKnowledgeGraphproject(BritishMuseum),http://www.researchspace.org

RichardsJ.,TudhopeD.&VlachidisA.(2015):TextMininginArchaeology:ExtractingInformationfromArchaeologicalReports,pp.240-254,in:BarceloJ.&BogdanovicI.(eds.):MathematicsinArchaeology.CRCPress;preprint,https://pure.york.ac.uk/portal/en/publications/text-mining-in-archaeology-extracting-information-from-archaeological-reports%28ef5831ea-4a00-4996-b225-ba53cf9019cf%29.html

Richards,Julian(2006):Archaeology,e-publicationandtheSemanticWeb.In:Antiquity,80(310):970-979,http://core.ac.uk/download/pdf/50930.pdf

RightField-SemanticdataannotationbyStealth,http://www.rightfield.org.uk

RodriguezEchavarriaK.,TheodoridouM.,GeorgisC.etal.(2012):Semanticallyrich3Ddocumentationforthepreservationoftangibleheritage.In:VAST’12-13thInternationalSymposiumonVirtualReality,ArchaeologyandCulturalHeritage,Brighton,UK,http://culturalinformatics.org.uk/files/papers/SemanticallyRich3D2012.pdf

Romanello,Matteo(2012):SKOSifyinganArchaeologicalThesaurus.In:ComputersfortheClassesweblog,8October2012,https://c4tc.wordpress.com/2012/10/08/skosifying-an-archaeological-thesaurus/

Romanello,Matteo(2014):MiningCitations,LinkingTexts.ISAWPaper7.24,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

RonzinoP.,AmicoN.,FelicettiA.&NiccolucciF.(2013):EuropeanstandardsforthedocumentationofhistoricbuildingsandtheirrelationshipwithCIDOC-CRM,pp.70-79,in:CRMEX2013–Workshop:PracticalExperienceswithCIDOCCRManditsExtensions,co-locatedwithTPDL2013,Valetta,Malta,26September2013,http://ceur-ws.org/Vol-1117/CRMEX2013.pdf

RonzinoP.,NiccolucciF.,FelicettiA.&DoerrM.(2016):CRMba,aCRMextensionforthedocumentationofstandingbuildings.In:InternationalJournalonDigitalLibraries,17(1):71-78,http://link.springer.com/article/10.1007%2Fs00799-015-0160-4

Ronzino,Paola(2015):CIDOCCRMba–ACRMextensionforbuildingarchaeologyinformationmodelling.PresentationatCIDOC-CRMSIG,32ndjointmeeting,OxfordUniversitye-ResearchCentre,11February2015,http://www.cidoc-crm.org/docs/32nd-meeting-presentations/CRMBA_Paola%20Ronzino_32SIG.pdf

Ronzino,Paola(2015):CIDOCCRMba:ACRMextensionforbuildingsarchaeologyinformationmodelling.UnpublishedPhDthesis,TheCyprusInstitute,Cyprus,January2015

RossS.,Ballsun-StantonB.,SobotkovaA.&CrookP.(2015):BuildingtheBazaar:EnhancingArchaeologicalFieldRecordingThroughanOpenSourceApproach,pp.111-129,in:WilsonA.T.&EdwardsB.(eds.):OpenSourceArchaeology:EthicsandPractice.WalterdeGruyter,



https://www.degruyter.com/downloadpdf/books/9783110440171/9783110440171-009/9783110440171-009.xml

RossS.,SobotkovaA.,Ballsun-StantonB,&CrookP.(2013):CreatingeResearchToolsforArchaeologists:TheFederatedArchaeologicalInformationManagementSystemsproject.In:AustralianArchaeology,No.77,December2013,https://www.academia.edu/5690498/Creating_eResearch_Tools_for_Archaeologists_The_Federated_Archaeological_Information_Management_Systems_project

Ross,Seamus(2003):PositionPaper,pp.7-11,in:DigiCULTThematicIssue3:TowardsaSemanticWebforHeritageResources.Salzburg,May2003,http://www.digicult.info/downloads/thematic_issue_3_low.pdf

Ross,Shawn(2015):CreatingInteroperableDigitalDatasets:theFederatedArchaeologicalInformationManagementSystems(FAIMS)Project.PresentationatMobilizingthePastforaDigitalFuture:thePotentialofDigitalArchaeology,WentworthInstituteofTechnology,Boston,27-28February2015,http://uwm.edu/mobilizing-the-past/sample-page-2/

RouechéC.,LawrenceK.&LawrenceK.F.(2014):LinkedDataandAncientWisdom.ISAWPaper7.25,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

SahooS.,HalbW.,HellmannS.etal.(2009):ASurveyofCurrentApproachesforMappingofRelationalDatabasestoRDF.W3CRDB2RDFIncubatorGroup,W3C,2009.http://esw.w3.org/Rdb2RdfXG/StateOfTheArt

Samwald,Matthias(2010):Commentsto“WhyCarrytheCostofLinkedData?”.TomHeathweblog,17June2010,http://tomheath.com/blog/2010/06/why-carry-the-cost-of-linked-data/

SchaibleJ.,GottronT.&ScherpA.(2014):ExtendedDescriptionoftheSurveyonCommonStrategiesofVocabularyReuseinLinkedOpenDataModeling.UniversitätKoblenz-Landau,ArbeitsberichteausdemFachbereichInformatik,Nr.1/2014,http://www.uni-koblenz.de/~fb4reports/2014/2014_01_Arbeitsberichte.pdf

Scheidel,Walter(2015):ORBIS:theStanfordgeospatialnetworkmodeloftheRomanworld.Princeton/StanfordWorkingPapersinClassics,May2015,http://orbis.stanford.edu/assets/Scheidel_64.pdf

SchmachtenbergM.,BizerC.&PaulheimH.(2014a):StateoftheLODCloud2014,Version0.4,30August2014,http://linkeddatacatalog.dws.informatik.uni-mannheim.de/state/

SchmachtenbergM.,BizerC.&PaulheimH.(2014b):AdoptionoftheLinkedDataBestPracticesinDifferentTopicalDomains,pp.245-260,in:TheSemanticWeb–ISWC2014.LectureNotesinComputerScience8796,http://dws.informatik.uni-mannheim.de/fileadmin/lehrstuehle/ki/pub/SchmachtenbergBizerPaulheim-AdoptionOfLinkedDataBestPractices.pdf

SchröttnerM.,HavemannS.,TheodoridouM.etal.(2012):Agenericapproachforgeneratingculturalheritagemetadata.4thInternationalEuro-MediterraneanConferenceonDigitalHeritage(EuroMed),Limassol,Cyprus,October2012,SpringerLNCS;https://www.semanticscholar.org/paper/A-Generic-Approach-for-Generating-Cultural-Schr%C3%B6ttner-Havemann/9e8d6f5201f153e4c03e066745967734a8fb5c2c

SebastianCuyS.,SchmidleW.&ThieryF.(2016):Linkingperiods:Modelingandutilizingspatio-temporalconceptsinthechronOntologyproject.PresentationatCAA2016Oslo,session“LinkedPasts:ConnectingIslandsofContent”,30March2016,



https://www.academia.edu/24845165/Linking_periods_Modeling_and_utilizing_spatio-temporal_concepts_in_the_chronOntology_project

SegersR.,VanErpM.,vanderMeijL.etal.(2011):Hackinghistory:Automatichistoricaleventextractionforenrichingculturalheritagemultimediacollections.Proceedingsofthe6thInternationalConferenceonKnowledgeCapture(K-CAP’11),http://ceur-ws.org/Vol-779/derive2011_submission_18.pdf

Seifreid,Rebecca(2014):LinkedOpenDatafortheUninitiated.ISAWPaper7.26,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

SemanticComputingResearchGroup(SeCo),AaltoUniversity,Finland,http://seco.cs.aalto.fi

Semanticweb.org:ListofSemanticAnnotationtools,http://semanticweb.org/wiki/Category:Semantic_annotation_tool

Semanticweb.org:SemanticWikiprojects,http://semanticweb.org/wiki/Semantic_wiki_projects

SEMIC-SemanticInteroperabilityCommunity,https://joinup.ec.europa.eu/community/semic/description

SEMLIB-SemanticToolsforDigitalLibraries(EUFP7-SMEproject),http://www.semlibproject.eu

SemWebQuality.org(providesinformationandtoolsaboutdataqualityinSemanticWebarchitectures),http://semwebquality.org

SENESCHAL-SemanticEnrichmentEnablingSustainabilityofArchaeologicalLinks(UKAHRC-fundedproject,2013-2014),http://hypermedia.research.glam.ac.uk/kos/SENESCHAL/;seealso:http://www.heritagedata.org/blog/about-heritage-data/seneschal/

ShadboltN.,Berners-LeeT.&HallW.(2006):TheSemanticWebRevisited.IEEEIntelligentSystems,vol.21,no.3,pp.96-101,http://eprints.soton.ac.uk/262614/1/Semantic_Web_Revisted.pdf

SibilledeGrimoüard,Claire(2014):ArchivesandLinkedData:Areourtoolsreadyto‘completethepicture’?Girona2014:ArxiusiIndústriesCulturals,Girona,Spain,11-15October2014,http://www.girona.cat/web/ica2014/ponents/textos/id9.pdf

Signore,Oreste(2009):Representingknowledgeinarchaeology:fromcataloguingcardstosemanticweb.In:ArcheologiaeCalcolatori,no.20,111-128,http://soi.cnr.it/archcalc/indice/PDF20/10_Signore.pdf

SimonR.,BarkerE.,deSotoP.&IsaksenL.(2014):Pelagios.ISAWPaper7.27,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

SimonR.,BarkerE.,IsaksenL.&deSotoCañamaresP.(2015):LinkingEarlyGeospatialDocuments,OnePlaceataTime:AnnotationofGeographicDocumentswithRecogito.In:e-Perimetron,10(2):49-59,http://oro.open.ac.uk/43613/1/Simon_et_al.pdf

SimonR.,HaslhoferB.&JungJ.(2011):Annotations,Tags&LinkedData-MetadataEnrichmentinOnlineMapCollectionsthroughVolunteer-ContributedInformation.6thInternationalWorkshoponDigitalApproachesinCartographicHeritageTheHague,Netherlands,7-8April2011,http://eprints.cs.univie.ac.at/2849/1/Simon_et_al._-_CartoHeritage_2011.pdf

SimonR.,IsaksenL.,BarkerE.&deSotoCañamaresP.(2016a):Peripleo:aToolforExploringHeterogeneousDatathroughtheDimensionsofSpaceandTime.In:Code4LibJournal,Issue31,http://journal.code4lib.org/articles/11144



SimonR.,IsaksenL.,BarkerE.&deSotoCañamaresP.(2016b):ThePleiadesGazetteerandthePelagiosProject.In:BermanM.L.,MosternR.&SouthallH.(eds.):PlacingNames:EnrichingandIntegratingGazetteers.IndianaUniversityPress,http://www.iupress.indiana.edu/product_info.php?cPath=1037_1116_3767&products_id=808056

SimovK.&KiryakovA.(2015):AccessingLinkedOpenDataviaaCommonOntology,pp.33-41,In:ProceedingsoftheSecondWorkshoponNaturalLanguageProcessingandLinkedOpenData,Hissar,Bulgaria,11September2015,https://aclweb.org/anthology/W/W15/W15-5506.pdf

SimperlE.,BürgerT.,HanglS.WörglS.&PopovI.(2012):ONTOCOM:AReliableCostEstimationMethodforOntologyDevelopmentProjects.In:JournalofWebSemantics,Vol.16,1-16;preprint,http://www.websemanticsjournal.org/index.php/ps/article/viewFile/320/320

Sinclair,P.A.S.etal.(2005):ConceptbrowsingformultimediaretrievalintheSCULPTEURproject.In:Proceedingsofthe2ndAnnualEuropeanSemanticWebConference,Heraklion,Crete,http://eprints.soton.ac.uk/260913/1/eswc.pdf

SITAR-SistemaInformativoTerritorialeArcheologicodiRoma,http://www.archeositarproject.it

SkevakisG.,MakrisK.,ArapiP.&ChristodoulakisS.(2013):ElevatingNaturalHistoryMuseums’CulturalCollectionstotheLinkedDataCloud.Proceedingsofthe3rdInternationalWorkshoponSemanticDigitalArchives(SDA),inconjunctionwithTPDL2013,http://ceur-ws.org/Vol-1091/paper4.pdf

Smith,MarcusJ.(2015):TheDigitalArchaeologicalWorkflow:ACaseStudyfromSweden,pp.215-220,in:CAA2014Paris-Proceedingsofthe42ndAnnualConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology,Paris,France,22-25April2014,Archaeopress,http://www.archaeopress.com/ArchaeopressShop/Public/download.asp?id={5CACE285-4C48-41AE-809E-E98B65C9E4CD}

Smith-Yoshimura,Karen(2014a):LinkedDataSurveyresults1–Who’sdoingit.In:Hangingtogether.orgOCLCResearchweblog,4September2014,http://hangingtogether.org/?p=4137

Smith-Yoshimura,Karen(2014b):LinkedDataSurveyresults2–Examplesinproduction.In:Hangingtogether.orgOCLCResearchweblog,4September2014,http://hangingtogether.org/?p=4147

Smith-Yoshimura,Karen(2014c):LinkedDataSurveyresults3–Whyandwhatinstitutionsareconsuming.In:Hangingtogether.orgOCLCResearchweblog,4September2014,http://hangingtogether.org/?p=4155

Smith-Yoshimura,Karen(2014d):LinkedDataSurveyresults4–Whyandwhatinstitutionsarepublishing.In:Hangingtogether.orgOCLCResearchweblog,4September2014,http://hangingtogether.org/?p=4167

Smith-Yoshimura,Karen(2014e):LinkedDataSurveyresults5–Technicaldetails.In:Hangingtogether.orgOCLCResearchweblog,4September2014,http://hangingtogether.org/?p=4256

Smith-Yoshimura,Karen(2014f):LinkedDataSurveyresults6-Advicefromtheimplementers.In:Hangingtogether.orgOCLCResearchweblog,4September2014,http://hangingtogether.org/?p=4284



Smith-Yoshimura,Karen(2014g):LinkedDataSurveyresults(resultsspreadsheet),https://groups.google.com/forum/#!topic/lod-lam/9ZR1FUvPntM

Smith-Yoshimura,Karen(2015):ResultsofLinkedDataSurveysforImplementers.Responses2014and2015(datasheet),http://oc.lc/0bglX7

Smith-Yoshimura,Karen(2016):AnalysisofInternationalLinkedDataSurveyforImplementers.In:D-LibMagazine,22(7/8),http://dx.doi.org/10.1045/july2016-smith-yoshimura

SNAC-SocialNetworksandArchivalContextproject(USA,2010-ongoing,InstituteforAdvancedTechnologyintheHumanities,UniversityofVirginia),http://socialarchive.iath.virginia.edu

SNAP-StandardsforNetworkingAncientProsopographies(UK,AHRCfundedproject,2014-2015),http://snapdrgn.net

Solanki,Monika(2009):SemanticwebinCulturalHeritageandArchaeology.W3CSemanticWeb,TracingNetworksWorkshop2009,UniversityofLeicester,13November2009,http://de.slideshare.net/nimonika/semantic-web-in-cultural-heritage-and-archaeology

SouzaR.,AlmeidaM.B.&TudhopeD.(2010):TheKOSspectra:atentativetypologyofKnowledgeOrganizationSystems.ISKO2010conference,Rome,23-26February2010,http://mba.eci.ufmg.br/downloads/ISKO%20Rome%202010%20submitted.pdf

SouzaR.,TudhopeD.&AlmeidaM.B.(2012):TowardsataxonomyofKOS:dimensionsforclassifyingknowledgeorganizationsystems.In:KnowledgeOrganization,39(3):179-192;preprint,http://mba.eci.ufmg.br/downloads/Souza_Tudhope_Almeida_-_KOS_Taxonomy.Submitted.pdf

SpampinatoD.&ZangaraI.(2013):ClassicalAntiquityandSemanticContentManagementonLinkedOpenData.In:1stInternationalWorkshoponCollaborativeAnnotationsinSharedEnvironment:Metadata,VocabulariesandTechniquesintheDigitalHumanities,Florence,10September2013(presentation),http://www.cs.unibo.it/dh-case/pdf/Zangara.pdf

StadlerC.,LehmannJ.,HöffnerK.&AuerS.(2012):LinkedGeoData:ACoreforaWebofSpatialOpenData.In:SemanticWebJournal,3(4):333-354http://jens-lehmann.org/files/2012/linkedgeodata2.pdf

STAR-SemanticTechnologiesforArchaeologicalResources(UK,AHRC-fundedproject,2007-2010),http://hypermedia.research.southwales.ac.uk/kos/star/

STELLAR-SemanticTechnologiesEnhancingLinksandLinkedDataforArchaeologicalResourcesproject(UK,AHRC-fundedproject,2010-2011),http://hypermedia.research.southwales.ac.uk/kos/stellar/

STELLARApplications(HypermediaResearchUnit,UniversityofSouthWales),http://hypermedia.research.southwales.ac.uk/resources/STELLAR-applications/

StevensonM.,OtegiA.etal.(2013):SemanticEnrichmentofCulturalHeritageContentinPATHS.PATHSproject,http://www.paths-project.eu/eng/content/download/5102/38896/file/SemanticEnrichment.pdf

Stevenson,Jane(2011):PuttingtheCaseforLinkedData.LOCAHProjectweblog,12July2011,http://locah.archiveshub.ac.uk/2011/07/12/putting-the-case-for-linked-data/

Stevenson,Jane(2012)LinkingLives:CreatingAnEnd-UserInterfaceUsingLinkedData.In:InformationStandardsQuarterly,24(2/3):14-23,http://dx.doi.org/10.3789/isqv24n2-3.2012.03



StuderR.&SureY.(2006):CostEstimationinOntologyEngineering.IST,Helsinki,November22,2006,slide7,ftp://ftp.cordis.europa.eu/pub/ist/docs/kct/cost-estimation-in-ontology-engineering_en.pdf

SuominenO.,PessalaS.,TuominenJ.etal.(2014):DeployingNationalOntologyServices:FromONKItoFinto.In:ISWC2014-13thInternationalSemanticWebConference,IndustryTrack,RivadelGarda,Italy,http://ceur-ws.org/Vol-1383/paper6.pdf

SwedishNationalHeritageBoard(2014):Listamedlämningstyperochrekommenderadantikvariskbedömning.Version4.1,2014-06-26,http://www.raa.se/app/uploads/2014/07/L%C3%A4mningstypslistan_ver-4_1_20140626.pdf

SwedishOpenCulturalHeritage(K-samsök):http://www.ksamsok.se/in-english/

Szabados,Anne-Violaine(2014):FromtheLIMCVocabularytoLOD.CurrentandExpectedUsesoftheMultilingualThesaurusTheA,pp.51-67,in:OrlandiS.etal.(2014):InformationTechnologiesforEpigraphyandCulturalHeritage.ProceedingsoftheFirstEAGLEInternationalConference,Paris,http://www.eagle-network.eu/wp-content/uploads/2015/01/Paris-Conference-Proceedings.pdf

SzekelyP.,KnoblockC.A.,YangF.etal.(2013):ConnectingtheSmithsonianAmericanArtMuseumtotheLinkedDataCloud.ESWC2013(LNCS7882,Springer),593-607,http://www.isi.edu/~szekely/contents/papers/2013/eswc-2013-saam.pdf

Taylor,Jon(2014):LinkeddataandthefutureofcuneiformresearchattheBritishMuseum.ISAWPaper7.28,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

TDWG-BiodiversityInformationStandards,http://www.tdwg.org

TEI-TextEncodingInitiative,http://www.tei-c.org/index.xml

ThieryF.&EngelT.(2016):TheLabelingSystem:Abottom-upapproachforenrichedvocabulariesinthehumanities,pp.259-268,in:CAA2015Siena-Proceedingsofthe43rdAnnualConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology.Oxford:Archaeopress,http://archaeopress.com/ArchaeopressShop/Public/download.asp?id={77DEDD4E-DE8F-43A4-B115-ABE0BB038DA7}

Thiery,Florian(2014):Linkingpotter,potsandplaces:aLODapproachtosamianware.PosterpresentedatCAA2014Paris,https://www.academia.edu/6782320/Linking_potter_pots_and_places_a_LOD_approach_to_samian_ware

Todorov,Ilian(2012):IstheWorkofScientificSoftwareEngineersRecognisedinAcademia?In:SoftwareSustainabilityInstituteweblog,http://software.ac.uk/blog/2012-04-23-work-scientific-software-engineers-recognised-academia

TolleK.&Wigg-WolfD.(2016):HowToMovefromRelationalto5StarLinkedOpenData–ANumismaticExample,pp.275-281,in:CAA2015Siena-Proceedingsofthe43rdAnnualConferenceonComputerApplicationsandQuantitativeMethodsinArchaeology,Volume1,Oxford:Archaeopress,http://archaeopress.com/ArchaeopressShop/Public/download.asp?id={77DEDD4E-DE8F-43A4-B115-ABE0BB038DA7}

Toms,ElaineG.(2015):ComplexToolsforComplexTasks.In:ProceedingsoftheFirstInternationalWorkshoponSupportingComplexSearchTasks(SCST2015),Vienna,Austria,29March2015.CEURWorkshopProceedings1338,http://ceur-ws.org/Vol-1338/paper_8.pdf



TounsiM.,FaronZuckerC.,ZuckerA.,VillataS.&CabrioE.(2015):StudyingtheHistoryofPre-ModernZoologywithLinkedDataandVocabularies,pp.7-14,in:SWASH2016-1stWorkshoponSemanticWebforScientificHeritage,Portoroz,Slovenia,1June2015,http://ceur-ws.org/Vol-1364/sw4sh-2015.pdf

TreeofLife(TOL)project,http://tolweb.org/tree/

Tsonev,Tsoni(2014):IntegratingHistorical-GeographicWeb-Resources.ISAWPaper7.29,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

TudhopeD.,BindingC.,JeffreyS.,MayK.&VlachidisA.(2011a):ASTELLARroleforknowledgeorganisationsystemsindigitalarchaeology.ASIS&TBulletin,37(4):15-18,http://www.asis.org/Bulletin/Apr-11/AprMay11_Tudhope_etAl.pdf

TudhopeD.,BindingC.,MayK.&CharnoM.(2013):PatternbasedmappingandextractionviatheCRM(-EH),pp.23-36,in:CRMEX2013–Workshop:PracticalExperienceswithCIDOCCRManditsExtensions,co-locatedwithTPDL2013,Valetta,Malta,26September2013,http://ceur-ws.org/Vol-1117/CRMEX2013.pdf

TudhopeD.,MayK.,BindingC.&VlachidisA.(2011b):Connectingarchaeologicaldataandgreyliteratureviasemanticcrosssearch.InternetArchaeology,Issue30,http://intarch.ac.uk/journal/issue30/tudhope_index.html

TzompanakiK.&DoerrM.(2012):FundamentalcategoriesandrelationshipsforintuitivequeryingCIDOC-CRMbasedrepositories.TechnicalReportICS-FORTH/TR-429,April2012,http://www.cidoc-crm.org/docs/TechnicalReport429_April2012.pdf

UBERON–UberAnatomyOntology,http://uberon.org;seealso:https://bioportal.bioontology.org/ontologies/UBERON

UnsworthJ.(2000):ScholarlyPrimitives:WhatMethodsDoHumanitiesResearchersHaveinCommon,andHowMightOurToolsReflectThis?SymposiumonHumanitiesComputing:formalmethods,experimentalpractice.King'sCollege,London,13May2000,http://people.brandeis.edu/~unsworth/Kings.5-00/primitives.html

Unsworth,John(2002):WhatisHumanitiesComputingandWhatisnot?In:ForumComputerphilologie,8November2002,http://computerphilologie.uni-muenchen.de/jg02/unsworth.html

vandeSompelH.,LagozeC.,NelsonM.L.etal.(2009):Addinge-scienceassetstothedataweb.LinkedDataontheWeb(LDOW2009),Madrid,Spain,20April2009,http://events.linkeddata.org/ldow2009/papers/ldow2009_paper8.pdf;seealsoarXiv:0906.2135v1[cs.DL],http://arxiv.org/abs/0906.2135

vanderMeijL.,IsaacA.&ZinnC.(2010):Aweb-basedrepositoryserviceforvocabulariesandalignmentsintheculturalheritagedomain.Proceedingsofthe7thEuropeanSemanticWebConference,Heraklion,Greece,30May-3June2010,394–409,http://www.few.vu.nl/~AI.Isaac/papers/STITCH-Repository-ESWC10.pdf

vanErpM.,OomenJ.,SegersR.etal.(2011):Automaticheritagemetadataenrichmentwithhistoricevents.ProceedingsofMuseumsandtheWeb2011,http://www.museumsandtheweb.com/mw2011/papers/automatic_heritage_metadata_enrichment_with_hi

vanHoolandS.&VerborghR.(2014):LinkedDataforLibraries,ArchivesandMuseums.Howtoclean,linkandpublishyourmetadata.FacetPublishing,http://book.freeyourmetadata.org



vanHoolandS.,DeWildeM.,VerborghR.,SteinerT.&VandeWalleR.(2015):ExploringEntityRecognitionandDisambiguationforCulturalHeritageCollections?In:LiteraryandLinguisticsComputing,30(2):262-279;preprint,http://freeyourmetadata.org/publications/named-entity-recognition.pdf

vanHoolandS.,VerborghR.&VandeWalleR.(2012a):JoiningtheLinkedDataCloudinaCost-EffectiveManner.In:InformationStandardsQuarterly,24(2/3):24-28,http://www.niso.org/apps/group_public/download.php/9423/IP_VanHooland-etal_%20LD-Cloud_isqv24no2-3.pdf

vanHoolandS.,VerborghR.,DeWildeM.,HercherJ.,MannensE.&VandeWalleR.(2012b):Evaluatingthesuccessofvocabularyreconciliationforculturalheritagecollections.In:JournaloftheAmericanSocietyforInformationScienceandTechnology,Vol.64:464–479;authors’paper,May2012,http://freeyourmetadata.org/publications/freeyourmetadata.pdf

VanKeer,Ellen(2014):MovingfromCross-CollectionIntegrationtoExplorationsofLinkedDataPracticesintheLibraryofAntiquityattheRoyalMuseumsofArtandHistory,Brussels.ISAWPaper7.30,http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/

vanOssenbruggenJ.,HildebrandM.&deBoerV.(2011):Interactivevocabularyalignment.TPDL2011-InternationalConferenceonTheoryandPracticeofDigitalLibraries,Berlin,Germany,26-28September2011,http://semanticweb.cs.vu.nl/lod/tpdl2011/paper.pdf(seealsotheusecasereplicabilitydocumentationhere:http://semanticweb.cs.vu.nl/lod/tpdl2011/)

VandenbusscheP.-Y.,AtemezingG.A.,Poveda-VillalónM.&VatantB.(2015):LinkedOpenVocabularies(LOV):agatewaytoreusablesemanticvocabulariesontheWeb.In:SemanticWebJournal,version29/09/2015,http://www.semantic-web-journal.net/system/files/swj1178.pdf

Vatant,Bernard(2012):Isyourlinkeddatavocabulary5-star?In:BernardVatantweblog,10February2012,http://bvatant.blogspot.fr/2012/02/is-your-linked-data-vocabulary-5-star_9588.html

VavliakisK.N.,KaragiannisG.T.&MitkasP.A.(2012):SemanticWebinCulturalHeritageafter2020.Workshopon“WhatwilltheSemanticWeblooklike10yearsfromnow?”heldinconjunctionwiththe11thInternationalSemanticWebConference2012(ISWC2012),Boston,USA,11November2012,http://stko.geog.ucsb.edu/sw2022/sw2022_paper10.pdf

VencesM.,GuayasaminJ.M.,MirallesA.&DelaRivaI.(2013):Tonameornottoname:CriteriatopromoteeconomyofchangeinLinnaeanclassificationschemes.In:Zootaxa,3636(2):201–244,http://biotaxa.org/Zootaxa/article/view/zootaxa.3636.2.1/1556

VIAF-VirtualInternationalAuthorityFile,http://viaf.org

Vici.org-ArchaeologicalAtlasofAntiquity,http://vici.org

Villazón-TerrazasB.&CorchoO.(2011):MethodologicalGuidelinesforPublishingLinkedData.OntologyEngineeringGroup,ComputerScienceSchool,PolytechnicUniversityofMadrid,http://delicias.dia.fi.upm.es/wiki/images/7/7a/07_MGLD.pdf

VlachidisA.&TudhopeD.(2011):SemanticAnnotationforIndexingArchaeologicalContext:APrototypeDevelopmentandEvaluation.In:MetadataandSemanticResearch(CommunicationsinComputerandInformationScience,Vol.240):363-374;preprint,http://hypermedia.research.glam.ac.uk/media/files/documents/2011-10-26/MTSR2011_Vlachidis_A-SemanticAnnoations-Camera_Ready.pdf



VlachidisA.&TudhopeD.(2013a):ClassicalArtSemanticsInformationExtraction:CASIEPilotProject.ConferenceoftheBritishChapteroftheInternationalSocietyforKnowledgeOrganization(ISKOUK2013),London,http://www.iskouk.org/conf2013/papers/VlachidisPaper.pdf

VlachidisA.&TudhopeD.(2013b):TheSemanticsofNegationDetectioninArchaeologicalGreyLiterature,pp.188-200,in:GaroufallouE.&GreenbergJ.(eds.):MetadataandSemanticsResearchCommunicationsinComputerandInformationScience,Vol.390;preprint,http://hypermedia.research.glam.ac.uk/media/files/documents/2015-04-28/The_Semantics_of_Negation_Detection_Camera_Ready.pdf

VlachidisA.&TudhopeD.(2015a):Aknowledge-basedapproachtoInformationExtractionforsemanticinteroperabilityinthearchaeologydomain.In:JournaloftheAssociationforInformationScienceandTechnology,67(5):1138-52,http://onlinelibrary.wiley.com/doi/10.1002/asi.23485/abstract

VlachidisA.&TudhopeD.(2015b):Negationdetectionandwordsensedisambiguationindigitalarchaeologyreportsforthepurposesofsemanticannotation.Program:electroniclibraryandinformationsystems,49(2):118-134,http://www.emeraldinsight.com/doi/abs/10.1108/PROG-10-2014-0076

VlachidisA.,BindingC.,MayK.&TudhopeD.(2010):Excavatinggreyliterature:acasestudyontherichindexingofarchaeologicaldocumentsviaNaturalLanguageProcessingtechniquesandknowledgebasedresources.In:ASLIBProceedings,62(4&5):466-475;preprint,http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.551.1066&rep=rep1&type=pdf

VlachidisA.,BindingC.,MayK.&TudhopeD.(2013):AutomaticMetadataGenerationinanArchaeologicalDigitalLibrary:SemanticAnnotationofGreyLiterature,pp.187-202,in:Przepiórkowski,Adametal.(eds.):ComputationalLinguistics–StudiesinComputationalIntelligence458.Springer;preprint,http://hypermedia.research.glam.ac.uk/media/files/documents/2011-11-02/Automatic_Metadata_Generation.pdf

Vlachidis,Andreas(2012):SemanticIndexingviaKnowledgeOrganizationSystems:ApplyingtheCIDOC-CRMtoArchaeologicalGreyLiterature.PhDThesis,UniversityofSouthWales,http://hypermedia.research.southwales.ac.uk/media/files/documents/2013-07-11/Andreas-Vlachidis_Thesis_print_ready.pdf

VOAF-VocabularyofaFriend,http://lov.okfn.org/vocommons/voaf/v2.3/

VocabularyMappingFramework(VMF),http://www.doi.org/VMF/

VocabularyMatchingTool(HypermediaResearchGroup,UniversityofSouthWales,UK),http://heritagedata.org/vocabularyMatchingTool/;sourcecodeforlocaldownloadandinstallation,https://github.com/cbinding/VocabularyMatchingTool

W3C(2001-2013)SemanticWebActivity,http://www.w3.org/2001/sw/

W3C(2004)Recommendation:ArchitectureoftheWorldWideWeb(Volume1),15December2004,http://www.w3.org/TR/webarch/#identification

W3C(2008)InterestGroupNote:CoolURIsfortheSemanticWeb,3December2008,http://www.w3.org/TR/cooluris/

W3C(2008)WorkingGroupNote:BestPracticeRecipesforPublishingRDFVocabularies,28August2008,https://www.w3.org/TR/swbp-vocab-pub/



W3C(2009)Recommendation:SimpleKnowledgeOrganizationSystem(SKOS)-Reference,18August2009,http://www.w3.org/2004/02/skos/

W3C(2011)InterestGroupNote:DescribingLinkedDatasetswiththeVoIDVocabulary,3March2011,http://www.w3.org/TR/void/

W3C(2012)Recommendation:OWL2-WebOntologyLanguageDocument-Overview(SecondEdition),11December2012,https://www.w3.org/TR/2012/REC-owl2-overview-20121211/

W3C(2012):OWL-WebOntologyLanguage–Currentstatus,http://www.w3.org/standards/techs/owl#w3c_all

W3C(2013)Recommendation:SPARQL1.1FederatedQuery,21March2013,http://www.w3.org/TR/sparql11-federated-query/

W3C(2013)WorkingGroupNote:ADMS-AssetDescriptionMetadataSchema,1August2013,http://www.w3.org/TR/2013/NOTE-vocab-adms-20130801/

W3C(2013)WorkingGroupNote:RDFa1.1Primer:RichStructuredDataMarkupforWebDocuments(secondedition),22August2013,http://www.w3.org/TR/xhtml-rdfa-primer;seealso:http://rdfa.info

W3C(2013):SPARQL-CurrentStatus,http://www.w3.org/standards/techs/sparql#w3c_all

W3C(2013-ongoing)DataActivity-BuildingtheWebofData,https://www.w3.org/2013/data/

W3C(2014)Recommendation:DCAT-DataCatalogVocabulary,16January2014,http://www.w3.org/TR/vocab-dcat/

W3C(2014)Recommendation:RDF1.1ConceptsandAbstractSyntax,25February2014,https://www.w3.org/TR/rdf11-concepts/

W3C(2014)Recommendation:RDFSchema1.1,25February2014,http://www.w3.org/TR/rdf-schema/

W3C(2014)WorkingGroupNote:BestPracticesforPublishingLinkedData,9January2014,https://www.w3.org/TR/ld-bp/

W3C(2015)Editor’sDraft:DataontheWebBestPracticesUseCases&Requirements,27March2015,https://www.w3.org/TR/dwbp-ucr/

W3C(2015):ResourceDescriptionFramework(RDF)-CurrentStatus,http://www.w3.org/standards/techs/rdf#w3c_all

W3Cwebsite:ListofTaggingtools,http://www.w3.org/2001/sw/wiki/Category:Tagging

W3Cwebsite:SemanticWebtools(fulllist):http://www.w3.org/2001/sw/wiki/SemanticWebTools

W3Cwiki:ConvertertoRDF,http://www.w3.org/wiki/ConverterToRdf

W3Cwiki:Tools,http://www.w3.org/2001/sw/wiki/Tools

Wallis,Richard(2012):WhatIsYourData’sStarRating(s)?Dataliberate.com,18January2012,http://dataliberate.com/2012/01/what-is-your-datas-star-ratings/

WangS.,IsaacA.,SchlobachS.etal.(2012):Instance-basedSemanticInteroperabilityintheCulturalHeritage.SemanticWebJournal,3(1),SpecialIssueonSemanticWebandReasoningforCulturalHeritageandDigitalLibraries,pp.45-64,http://www.few.vu.nl/~AI.Isaac/publications.php



WellsJ.J.,KansaE.,YerkaS.J.etal.(2014):Web-baseddiscoveryandintegrationofarchaeologicalhistoricpropertiesinventorydata:TheDigitalIndexofNorthAmericanArchaeology(DINAA).In:LiteraryandLinguisticComputing,3(29):349-360;https://www.academia.edu/11450026/Web-based_discovery_and_integration_of_archaeological_historic_properties_inventory_data_The_Digital_Index_of_North_American_Archaeology_DINAA_

Wester,JeroenandNederbragt,Hans(2007):RNA-project:Usingthingslikethesauriandtaxonomiesinrealcases!,pp.93-99,in:Aroyo,L.,Hyvönen,E.andvanOssenbruggen,J.(2007):CulturalHeritageontheSemanticWeb.Workshop9ofthe6thInternationalSemanticWebConference,Korea,2007http://www.cs.vu.nl/~laroyo/CH-SW/ISWC-wp9-proceedings.pdf

Whitcher-Kansa,Sarah(2015):UsingLinkedOpenDatatoImproveDataReuseinZooarchaeology.In:EthnobiologyLetters,6(2):224-231,http://ojs.ethnobiology.org/index.php/ebl/article/view/467/254

WickettK.M.,IsaacA.,DoerrM.etal.(2014):RepresentingCulturalCollectionsinDigitalAggregationandExchangeEnvironments.In:D-LibMagazine,20(5-6),May/June2014,http://www.dlib.org/dlib/may14/wickett/05wickett.html

WiljesC.,JahnN.,LierF.etal.(2013):TowardsLinkedResearchData:AnInstitutionalApproach.3rdWorkshoponSemanticPublishing(SePublica),CEURWorkshopProceedings,Aachen:27–38,http://ceur-ws.org/Vol-994/paper-03.pdf

Wilson,Scott(2014):PreservingandCuratingSoftware.OSSWatchwebsite,guidancematerial,5November2014,http://oss-watch.ac.uk/resources/preservation

WolstencroftK.,OwenS.,HorridgeM.etal.(2011):RightField:Embeddingontologyannotationinspreadsheets.In:Bioinformatics27(14):2021-22,http://bioinformatics.oxfordjournals.org/content/27/14/2021.full

Wolstencroft,Katy(2012):RightField:SemanticEnrichmentofSystemsBiologyDatausingSpreadsheets(myGrid,SysMO-DB,UniversityofManchester).PresentationatIEEE-Escience2012,Chicago,USA,11October2012,https://seek.sysmo-db.org/presentations/61/download

WoodD.,ZaidmanM.,RuthL.withHausenbladM.(2014):LinkedData.StructuredDataontheWeb.ShelterIsland,NY:Manning,http://www.manning.com/dwood/

WorldGeodeticSystem1984(WGS84),http://earth-info.nga.mil/GandG/wgs84/

Wright,Holly(2011):SeeingTriple.Archaeology,FieldDrawingandtheSemanticWeb.PhDDissertation.TheUniversityofYork,DepartmentofArchaeology,September2011,http://etheses.whiterose.ac.uk/2194/1/WrightThesis.pdf

YuC.-H.(2010):SemanticAnnotationof3DDigitalRepresentationofCulturalArtefacts.BulletinofIEEETechnicalCommitteeonDigitalLibraries(TCDL),vol.6,issue.2,http://www.ieee-tcdl.org/Bulletin/v6n2/Yu/yu.html

Zaino,Jennifer(2013):Artloverswillseethere’smoretolovewithlinkeddata.Semanticweb.com,21June2013,https://semanticweb.com/art-lovers-will-see-theres-more-to-love-with-linked-data_b38088#more-38088

ZaveriA.,RulaA.,MaurinoA.,PietrobonR.,LehmannJ.&AuerS.(2013):QualityAssessmentforLinkedOpenData:ASurvey.SemanticWebJournal,556,http://www.semantic-web-journal.net/system/files/swj556.pdf



ZengM.L.&ŽumerM.(2013):AMetadataApplicationProfileforKOSVocabularyRegistries.ISKOUKBiennialConference:KnowledgeOrganization–pushingtheboundaries,London,8-9July2013,http://www.iskouk.org/sites/default/files/ZengPaper_1.pdf

ZengM.L.&ŽumerM.(2015):NetworkedKnowledgeOrganizationSystemsDublinCoreApplicationProfile(NKOSAP),2015-10-03,http://nkos.slis.kent.edu/nkos-ap.html

ZhangY.,OgletreeA.,GreenbergJ.&RowelC.(2015):ControlledVocabulariesforScientificData:UsersandDesiredFunctionalities.In:2015AnnualMeetingoftheAssociationforInformationScience&Technology,St.Louis,USA,6-10November2015;preprint,https://wakespace.lib.wfu.edu/bitstream/handle/10339/57209/zhang-ogletree-greenberg-rowell-controlled-vocabularies-for-scientific-data-preprint.pdf

Zimmermann,Antoine(2010):Ontologyrecommendationforthedatapublishers.ORES-2010-Proceedingsofthe1stWorkshoponOntologyRepositoriesandEditorsfortheSemanticWeb,Hersonissos,Crete,Greece,May31st,2010,http://ceur-ws.org/Vol-596/paper-12.pdf

ZOOMATHIA:Transmissionculturelledessavoirszoologiques(Antiquité-MoyenÂge):discoursettechniques,http://www.cepam.cnrs.fr/zoomathia/

ZuiderwijkA.,JefferyK.&JanssenM.(2012):Thepotentialofmetadataforlinkedopendataanditsvalueforusersandpublishers.In:JeDEM-eJournalofeDemocracyandOpenGovernment,4(2):222-244,http://www.jedem.org/index.php/jedem/article/view/138/113

Data & Analytics

ARIADNE: Report on the ARIADNE Linked Data Cloud