Embed Size (px)
Citation preview
M
NCa
b
c
d
a
ARRA
KLMIAA
1
nsbirialcRfgu2
lbdsou
0d
Computers and Electronics in Agriculture 70 (2010) 302–320
Contents lists available at ScienceDirect
Computers and Electronics in Agriculture
journa l homepage: www.e lsev ier .com/ locate /compag
etadata interoperability in agricultural learning repositories: An analysis
ikos Manouselis a,∗, Jehad Najjar c, Kostas Kastrantas a, Gauri Salokhe b,hristian M. Stracke d, Erik Duval c
Greek Research & Technology Network (GRNET S.A.), and Informatics Laboratory, Agricultural University of Athens, 75 Iera Odos Str., 11855 Athens, GreeceFood and Agriculture Organization of the United Nations, Viale delle Terme di Caracalla, 00153 Rome, ItalyComputer Science Dept., K.U. Leuven, Celestijnenlaan 200A, B-3001 Leuven, BelgiumInformation Systems for Production and Operations Management, University of Duisburg-Essen, Campus Essen, Universitaetsstr. 9, D-45141 Essen, Germany
r t i c l e i n f o
rticle history:eceived 15 December 2008eceived in revised form 30 May 2009
a b s t r a c t
The rapid evolution of ICT creates numerous opportunities for agricultural education and training. Dig-ital learning resources are organized in online databases called learning repositories, in which peoplecan search, locate, and access resources. In order to facilitate the exchange of information between such
ccepted 14 July 2009
eywords:earning repositoryetadata
nteroperability
repositories, the issue of metadata interoperability is crucial. In this paper, we particularly focus on meta-data interoperability of learning repositories with content relevant to agricultural stakeholders. Morespecifically, we present results from an analysis of implementations of metadata standards in agricul-tural learning repositories around the world. The results provide useful feedback to the developers ofrepositories with educational content for agricultural stakeholders, as well as directions for potential
this
pplication profilenalysisharmonization of work in
. Introduction
The rapid evolution of Information and Communication Tech-ologies (ICT) creates numerous opportunities for providing newtandards of quality in educational services. Internet increasinglyecomes a dominant medium for learning, training and work-
ng, and the amount of digitally (i.e. online) available learningesources is growing exponentially. Online learning resources maynclude courses, best practices, presentations, reports, textbooks,s well as other types of digital resources that can be used forearning purposes. They cover numerous didactical topics, such asomputing, business, art, engineering, technology and agriculture.esources are offered from various types of organizations, at dif-
erent cost rates, and targeting different categories of learners. Ineneral, the potential of Internet-available resources that can besed to facilitate learning and training is rapidly increasing (Friesen,001).
The digital resources that are developed to support teaching andearning activities have to be easily located and retrieved, as well ase suitably selected to meet the needs of those to whom they are
elivered. For this purpose, database systems that facilitate theirtorage, location and retrieval have been developed and deployednline (Holden, 2003). Such systems, termed as repositories, aresed to store any type of digital material. However, repositories for∗ Corresponding author. Tel.: +30 2107474267.E-mail address: [email protected] (N. Manouselis).
168-1699/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.compag.2009.07.007
area.© 2009 Elsevier B.V. All rights reserved.
learning resources are considerably more complex, both in terms ofwhat needs to be stored and how it may be delivered. The purposeof a repository with learning resources is not simply safe storageand delivery of the resources, but mainly the facilitation of theirreuse and sharing (Duncan, 2002). Therefore, the repositories thatare developed to provide access to digital learning resources aretermed as learning repositories or LRs (Holden, 2003).
There is a foreseen potential for agricultural stakeholders, fromhaving access to online learning resources and repositories. Forthem, the role of education (and lifelong learning in most cases)is crucial: for instance, better educated farmers of all ages andbackgrounds, with ample lifelong learning opportunities and accessto online learning resources, can resist urbanisation tendencies,protect the natural and human resources of the countryside, under-stand the new challenges for rural areas and respond to them withnew initiatives, flexibility and adaptability (Berge and Leary, 2006;Tzikopoulos and Yialouris, 2007). General-topic LRs do not seem tofocus particularly on the learning resources that are of interest toagricultural stakeholders (Tzikopoulos et al., 2005). This is the rea-son why a number of agricultural learning repositories (AgLRs) havebeen developed and deployed during the last few years (Manouselisand Salokhe, 2008). Nevertheless, recent studies have indicated thatthe implementation of such systems in the agricultural domain
is taking place in a widely dispersed manner (Manouselis et al.,2009; Manouselis and Salokhe, 2008). Similar applications oftenuse dissimilar data models, and this hampers information sharingand reuse between AgLRs. This practice leads to the lack of inter-operability among different implementations: that is, their ability
lectro
tf
osafebSawsbwWLF
baotpoyoStt6a
2
2
WSmocisab
Hwtiidibiaacic
oC“
N. Manouselis et al. / Computers and E
o cooperate and exchange information in pre-agreed, well-definedormats.
To this end, this paper addresses a particular type of inter-perability in AgLR systems, i.e. metadata interoperability. Morepecifically, the paper studies the way that metadata is specifiednd used in a number of different systems, and identifies pointsor further improvement and harmonization. To achieve this, itngages a set of principles and practicalities that have been definedy the Workshop on Learning Technologies (WS-LT) of the Europeantandardization Committee CEN (CWA 15555, 2006). The papernalyses existing solutions and implementations, identifies theireaknesses and challenges that have to be addressed, proposes
uggestions to the developers of such solutions, and outlines someasic directions for potential harmonization of similar efforts. Thisork has taken place in the context of an expert team of the CENS-LT (CWA 15966, 2009) with the support of the Agricultural
earning Repositories Task Force (AgLR-TF, http://aglr.aua.gr) of theood and Agriculture Organization of the United Nations (FAO).
The remainder of this paper is structured as follows. First, theackground of this study is presented in Section 2, by providingn introduction of the basic concepts related to metadata inter-perability, such as metadata application profiles. In Section 3,he methodology of this study is presented: The principles andracticalities for building metadata application profiles for LRs areutlined, and then the steps followed within this particular anal-sis are explained. Moreover, the analysis results from a samplef existing metadata application profiles in AgLRs are described inection 4. Based on this analysis, a number of outcomes are iden-ified in Section 5. A discussion on the basic outcomes, as well ashe challenges and limitations of this study is included in Section. Finally, in Section 7, the main conclusions of this study are madend further directions of work are identified.
. Background
.1. Learning objects
Long before the advent and wide adoption of the World Wideeb (WWW), researchers such as Ted Nelson (1965) and Roy
tringer (1992) referred to environments where the design of infor-ation and courses could be based on the notion of reusable
bjects. The idea of creating educational components from existingomponents rather than building those components from scratchs as old as, at least, the conceptual design of the Xanadu hypertextystem (Nelson, 1965). In Xanadu, each document may consist ofny number of parts each of which may be of any data type and cane referenced from any other document.
In 1994, the concept of learning objects was coined by Wayneodgins (Hodgins, 2002). It was inspired by LEGO® children toys,here different small LEGO® components are assembled together
o form new larger structures. The main idea behind the learn-ng object concept is that instructional designers can build smallnstructional components that can be reused and customized inifferent contexts (Wiley, 2000). In this way, instead of creating
nstructional material from scratch, instructional designers canuild teaching material by assembling and reusing available small
nstructional components (i.e. the learning objects). For instance,set of available images, text fragments, presentation slides and
udio tracks may be used to create, for example, a lesson or aourse. On the other hand, a lesson object may also be decomposednto smaller components that can be reused in other lessons and
ontexts.In this direction, one of the most popular definitions of a learningbject has been given by the IEEE Learning Technology Standardsommittee in the IEEE Learning Object Metadata (LOM) standard:. . .a learning object is defined as any entity, digital or non-digital,
nics in Agriculture 70 (2010) 302–320 303
that may be used for learning, education or training.” (IEEE LOM,2002). Wiley (2002) restricted this definition by characterising alearning object as “. . .any digital resource that can be reused tosupport learning”. Metros and Bennet (2002) noted that learningobjects should not be confused with information objects that haveno learning aim. It has also been argued (McCormick, 2003) thatlearning objects should include (either within or in a related doc-umentation) some learning objectives and outcomes, assessments,and other instructional components, as well as the object itself.
2.2. Metadata
In order to enable the discovery of learning objects, descrip-tive information (metadata) about each learning object needs to becreated. Metadata is simply defined as “data about data” or “infor-mation about information” (Miller, 1996; Steinacker et al., 2001;Taylor, 2003; NISO, 2004; Sen, 2004) and is structured informationthat identifies, describes, explains, locates, or otherwise makes iteasier to retrieve, use, or manage a resource. For instance, in thecontext of a library, metadata about a book may include is its title,author, publication date and some general description about itscontent. In the context of music, metadata about a musical trackmay include the title of the song, the artist, the album and therelease year. In the learning object context, metadata may includethe title of the object, a description of its content, the subject itcovers and maybe the type and name of the file. In LRs, metadataenables the discovery and reuse of the objects described. Based onthe instructional and other properties of the object (e.g. title, tar-get user group, subject domain or granularity), users may searchfor relevant learning objects. In addition, the metadata provide theusers with the information needed to decide whether an object isappropriate for (re)use in a particular task or context.
Metadata is made up of data items that are associated to theresource, the so-called metadata elements. Metadata schemas (ormetadata models) are sets of metadata elements designed for a spe-cific purpose, such as describing a particular type of resource (NISO,2004). Metadata specifications are well-defined and widely agreedmetadata schemas that are expected to be adopted by the majorityof implementers in a particular domain or industry. When a speci-fication is widely recognized and adopted by some standardizationorganization, it then becomes a metadata standard.
Despite the existence of numerous metadata standards, there isno all-encompassing one to be used in every application. Rather,there are various metadata standards or specifications that can beadapted or “profiled” to meet community context-specific needs(Kraan, 2003). This conclusion has lead to the emergence of theapplication profile concept. A metadata application profile (AP) isan assemblage of metadata elements selected from one or moremetadata schemas, and its purpose is to adapt or combine existingschemas into a package that is tailored to the functional require-ments of a particular application, while retaining interoperabilitywith the original base schemas (Roberts, 2003; Duval et al., 2002,2006; Heery, 2002).
Several schemas of metadata are available for indexing learningobjects. Standards are proposed for different purposes and con-texts. The Dublin Core Metadata Element Set (DCMES) (Dublin Core,2004) is a standard for describing general information resourcesavailable online. IEEE LOM (IEEE, 2002) is a standard particularlydeveloped for describing learning objects. The LOM standard isbased on early versions of the metadata sets of Alliance of RemoteInstructional Authoring and Distribution Network for Europe (ARI-
ADNE) (Duval et al., 2001) and Instructional Management Systems(IMS, 2007).When developing an LR, the first step to be taken is definingthe metadata that will be implemented to describe (annotate) thelearning objects to be stored. Metadata designers typically create
3 lectron
aCwonict(mta
2
ot2ireo
tstesr(oa
osto2c2itet
2
iooV2aRaeaa(aiie
04 N. Manouselis et al. / Computers and E
n application profile by customizing a base schema, such as Dublinore or IEEE LOM. They need to select the metadata elements thatill be included in the application profile, and this selection is based
n the needs and context of the served environment. To meet theeeds of the user audiences that the LR will serve, the base schema
s usually adapted in several ways. For instance, this may involvehanging an element’s obligation status (e.g. optional to manda-ory) or the addition of more metadata elements and vocabulariesvalues) than those contained in the base schema. Not all of these
odifications are acceptable (permitted), and this often leads tohe lack of interoperability between LRs that implement differentpplication profiles of the same base schema.
.3. Metadata interoperability
Interoperability has been judged essential in order to realize anpen infrastructure for learning that can put a wide diversity ofools and content in the hands of learners and tutors (PROLEARN,006). Interoperability issues are relevant for organizing learn-
ng resources into digital repositories, interconnecting learningepositories and exchanging metadata and queries, creating andxchanging open learning activities, as well as ensuring the qualityf the published content.
Interoperability itself has been defined in a number of ways. Inhe context of metadata, however, it usually refers to the ability of aystem to process metadata records produced by a third party sys-em (CWA 15555, 2006). This metadata exchange enables users inach repository, using compatible tools and services (e.g. federatedearch), to extend their access beyond only locally available mate-ial to a variety of learning objects collected in other repositoriesNajjar, 2008). This means that LR users can access a large numberf learning objects that may cover a variety of scientific domains,ge ranges and languages.
There are two levels of interoperability required for the exchangef metadata records. Structural interoperability ensures that thetructure of metadata instances produced by a repository conformso a common base schema (Duval et al., 2006), while semantic inter-perability concerns the meaning of the metadata records (Euzenat,002). Metadata records that conform to a base schema are pro-essed by services in different (but compatible) repositories (Heery,002). While designing the APs that different LRs use, the changes
mposed to the base schemas often lead to modifications that someimes break the interoperability rules and do not allow two differ-nt LRs to exchange metadata. In this paper, we examine if and howhis happens in the particular case of AgLRs.
.4. Metadata APs in LRs
There have been several studies investigating the usage andmplementation of metadata APs in various LRs, although onlyne covered AgLRs in particular. General studies examine vari-us aspects of learning objects and LRs (Pisik, 1997; Balanskat anduorikari, 2000; Neven and Duval, 2002; Haughey and Muirhead,004; Retalis, 2004; Riddy and Fill, 2004). Some of them includen LR review in their literature (e.g. Haughey and Muirhead, 2004;etalis, 2004), some focus on a particular LR segment (e.g. Nevennd Duval, 2002), some study the LR users and usage (e.g. Najjart al., 2003), and others have some particular geographical cover-ge (e.g. Balanskat and Vuorikari, 2000; EdReNe, 2008). There arelso studies examining how LRs are deployed and implemented
Tzikopoulos et al., 2007; Ochoa and Duval, 2009). Finally, therere some studies particularly focusing on the way metadata APsmplemented in LRs. A typical example is the study of metadatanteroperability in institutional repositories by Bueno-de-la-Fuentet al. (2009).ics in Agriculture 70 (2010) 302–320
Nevertheless, apart from a study that partially covered theimplementation of metadata standards in AgLRs (Tzikopoulos et al.,2005) and some recent comparison of two particular implementa-tions of metadata APs in AgLRs (Manouselis et al., 2009), there hasnot been a systematic analysis and comparison of the way metadataAPs are implemented in AgLRs, as well as how well they comply tothe rules and restrictions of the base metadata schemas. To this end,this paper particularly focuses on metadata implementations in LRsfor agriculture, food and the environment.
3. Methodology
3.1. Principles, practicalities and good practice
As described in the guidelines produced by CEN WS-LT (CWA15555, 2006), APs take one or more base schemas as their startingpoint and impose additional restrictions on it. Modifications of thiskind, limit the options available to a subset of those available in theoriginal base schema (e.g. reducing a vocabulary list). The goal is toincrease interoperability beyond the level of the base schema, whileretaining interoperability with those applications that are unawareof the rules of the particular application profile. For instance, a LRthat conforms to the IEEE LOM standard should be able to processa metadata instance from a LR that conforms to some IEEE LOMapplication profile. According to the guidelines (CWA 15555, 2006),what is not permitted are modifications to the base schema thatbreak its conformance rules.
The steps proposed for developing e-learning metadata APs inthe CEN WS-LT guidelines, are the following (CWA 15555, 2006):
(a) Start from own requirements: the basic goal of an AP is to supportspecific requirements of a particular context through a pro-file of a generic standard. It is important to have an explicitunderstanding of those specific requirements, in the form of aclear scope and purpose statement. A particularly effective wayto elicit requirements is the definition of so-called use cases thatdescribe how an end user would make use of the application tobe developed.
(b) Data elements:◦ Selection of data elements. Once the requirements are clari-
fied, a first important decision in the actual development ofa metadata AP is the selection of data elements that the APwill be built from. Often, the profile developers start from ametadata schema that has a scope and purpose similar to thatof the AP.
◦ Dealing with size and smallest permitted maximum. A smallestpermitted maximum (SPM) is the smallest number of occur-rences of a field that an application should support whenreading, writing or otherwise processing metadata instances.As a general rule, an application profile can reduce the sizeof a data element, or keep it equal to the value in the basestandard. An AP cannot increase the size of a data element.
◦ Data elements from multiple namespaces. In principle, an APcan be based on more than one base metadata schema. How-ever, there seems to be very little practice doing somethingsimilar for a particular metadata AP.
◦ Adding local data elements. Besides mixing and matching dataelements from several base standards, an AP may also includelocal data elements.
(c) Obligation of data elements: once the full set of metadata ele-ments to be included in the AP has been decided upon, the
status of these data elements can be defined (e.g. mandatory,conditional, recommended, optional). An AP can impose morestringent obligations on data elements than the base stan-dard does. An AP cannot relax such obligations: for instance,a mandatory element cannot lose its mandatory status in an AP.
N. Manouselis et al. / Computers and Electronics in Agriculture 70 (2010) 302–320 305
Table 1Overview of agricultural LRs operating during autumn 2008 (source: http://aglr.aua.gr).
Repository name URL Organization Country
Centre National de RechercheAgronomique (CNRA)
http://www.cnra.ci BOUAN Boumi Boniface Cote D’Ivoire
CGIAR On-line Learning Resources http://learning.cgiar.org Consultative Group on InternationalAgricultural Research (CGIAR)
United States
COTR’s e-training site http://kirk.estig.ipbeja.pt/cotr/ Centro Operativo e de Tecnologia de Regadio(COTR)
Portugal
EcoLearnIT http://ecolearnit.ifas.ufl.edu Soil and Water Science Department, Universityof Florida
United States
FAO Capacity Building Portal http://www.fao.org/capacitybuilding/ Food and Agriculture Organization of theUnited Nations
Italy
Lao Agriculture Database http://lad.nafri.org.la National Agriculture and Forestry ResearchInstitute of Lao PDR
Lao People’s DemocraticRepublic
Network of Aquaculture Centres inAsia-Pacific
http://www.enaca.org Network of Aquaculture Centres in Asia-Pacific Thailand
Rural-eGov Observatory http://rural-egov.eu Informatics Lab, Agricultural University ofAthens
Greece
SANREM CRSP Knowledge Base http://www.oired.vt.edu/sanremcrsp/menuinformation/SKB.php
SANREM CRSP United States
T
(
rbfAi
3
3
(sTtt
•
member of the expert team. An analysis report (including both
rAgLor - Turkish AgriculturalLearning Object Repository
http://traglor.cu.edu.tr
d) Value space: the value space defines the set of values that thedata element shall derive its value from. The AP may be morerestrictive about the value space of a data element than the basestandard is; it cannot be less restrictive.
(e) Relationship and dependency: more complex inter-relationshipsand dependencies between data elements can also be definedin an AP. The AP may be more restrictive about such inter-relationships than the base standard is; it cannot be lessrestrictive.
(f) Data type profiling: in effect, data types can be considered meta-data schemas in their own rights. Therefore, all the rules definedabove for APs of metadata schemas are also applicable to datatypes.
(g) Application profile binding: the general rule on the level of a bind-ing of an AP (e.g. in XML or RDF) is to make sure that any instancethat conforms to the relevant binding of the base standard alsoconforms to the binding of the AP. For instance, in XML bind-ings it is important to make sure that AP data element namesare either the names from the corresponding data elements inthe base standard, or declared explicitly as subclasses of thesedata elements.
These guidelines and recommendations that the CEN WS-LTeport provides to the developers of e-learning metadata APs cane easily translated into a checklist of points that have to be care-
ully addressed. Using such a checklist (like the one included inppendix A), an analysis of existing APs can take place, as we show
n the following paragraphs.
.2. Method of analysis
.2.1. Analysis toolBased on the guidelines and recommendations of CWA 15555
2006) and Najjar et al. (2004), a number of analysis dimen-ions have been incorporated into an appropriate analysis tool.his aimed at supporting the analysis of existing agricultural APs,hrough a template that has been developed as an Excel file. Theemplate included the following components:
An overall overview of the analysed AP, which includes generalinformation (such as its title, description, and producer), informa-tion about existing documentation (such as a conceptual modeland data bindings), information about its scope and purpose (such
Cukurova University, Faculty of Agriculture,Div. of Biometry & Genetics
Turkey
as a clear scope definition and use cases), and an overview of theresults of the mapping of the AP into its base schema (particularlyfocusing on allowed and non-allowed modifications).
• A detailed mapping of the analysed AP onto its base schema(s),i.e. IEEE LOM, DC, or both.
The tool is included in Appendix A of the paper.
3.2.2. CoverageA number of AgLRs have been examined, in order to identify the
metadata that they are using. From the AgLR Task Force listing thatis illustrated in Table 1 (source: http://aglr.aua.gr), as well as otherexternal sources, the following sample of agricultural APs has beenassembled:
1. Rural-eGov IEEE LOM AP (ReGov LOM).2. FAO Agricultural Learning Resources AP (FAO Ag-LR).3. CGIAR LOM Core AP (CG LOM Core).4. BIOAGRO LOM AP.5. Biosci Education Network (BEN) AP.6. Sustainable Agriculture and Natural Resource Management Col-
laborative Research Support Program (SANREM CRSP) AP.7. TrAgLor LOM AP.8. Intute: Health and Life Sciences AP (Intute AP).9. EcoLearnIT LOM AP.
From those mentioned above, it was possible to analyse in detailonly the first six. For the TrAgLor LOM AP only a preliminary analysistook place, based on basic information that was provided by the APdevelopers (such as a database instance of its implementation). Forthe Intute and EcoLearnIT LOM APs, no analysis was possible (theyare included here for reference reasons).
3.2.3. ProcessSix (6) experts have participated in the team that took over the
analysis. Each AP in the sample has been assigned to at least one
general recommendations as well as a mapping of the AP to itsbase schema) has been produced. Finally, all AP reports have beenintegrated into one overall report. The main results of the analy-sis are being presented in the following section (Table 2 provides acondensed view of the analysis reports).
306 N. Manouselis et al. / Computers and Electronics in Agriculture 70 (2010) 302–320
Table 2Results from the analysis of the metadata APs.
Title of AP ReGov LOM FAO Ag-LR CG LOM Core BIOAGROLOM
BEN SANREMCRSP
TrAgLorLOM
DocumentationConceptual data model Yes Yes Yes Yes Yes Yes NoTechnical bindings No XML No No No No XMLClaims conformance No Yes Yes No No No Yes
Scope and purposeClearly stated in documentation Yes Yes Yes Yes Yes Yes NoUse cases No Yes Yes No No No No
Selection of elementsAll mandatory selected Yes Yes Yes Yes Yes N/A NoAdditional elements No Yes No No No Yes YesIf additional, schema conformance Yes Yes NoElement selection Sub-set Multi-source Sub-set Sub-set Sub-set Sub-set and
Ad hocComplete
Value space selection Multi-source Multi-source Multi-source Ad hoc Multi-sourceand Ad hoc
Ad hoc –
Allowed modificationsMandatory selection of non-mandatory data elements Yes Yes Yes Yes Yes Yes –Changing/defining size and smallest permitted maximum Yes Yes No Yes No No –Change obligation of elements Yes Yes Yes Yes Yes Yes –Value space(s) modifications Yes Yes Yes Yes Yes No –Other modifications No No No Yes No No –
Non-allowed modificationsAltering the relative location of an existing data element No No No No No No –Creating a new element that mimican existing element No No No No No No –Changing the meaning of an existing element No No No No No No –Changing the name of an element No No No No No No –Extending a schema other than at a specified extension point No No No No No No –Making a mandatory element optional No No No No No No –
YesNoNoNo
4
A
4
LtdaeoStafiG
2ooV
4
cbott
Extending cardinality of an element NoAdding new items in a controlled vocabulary list NoBreaking base schema conformance rules NoOther modifications No
. Analysis of agricultural application profiles
In each paragraph, the analysis of the corresponding agriculturalP is presented.
.1. ReGov LOM
Rural-eGov IEEE LOM (or simply ReGov LOM) AP is an IEEEOM-based AP that has been developed to facilitate the descrip-ion and categorization of learning resources that have beeneveloped to support the training of rural and agricultural smallnd medium enterprises (SMEs) on topics related to the use of-government (Manouselis et al., 2007). Resources are being devel-ped for the SMEs in five rural areas of Europe: UK, Greece, Poland,lovenia and Germany. They are being populated in a reposi-ory of learning resources, called the Rural-eGov Observatory. Inddition, they are also described with metadata in one of theve targeted languages (i.e. English, Greek, Polish, Slovenian anderman).
The application profile has been published on 25 September007 by the Informatics Laboratory of the Agricultural Universityf Athens, in the context of the project “Rural-eGov: Training SMEsf Rural Areas on Using e-Government Services” of the Leonardo dainci Programme (http://rural-egov.eu).
.1.1. DocumentationA tree view of the conceptual schema of the ReGov LOM appli-
ation profile is illustrated in Fig. 1. As Table 2 shows, no technicalinding is provided for ReGov LOM, although it might be helpful forther developers. The implementation of ReGov LOM is in a rela-ional database, therefore no technical conformance is claimed forhe XML binding of LOM.
No Yes No No –Yes Yes No No –No Yes No No –No Yes Yes Yes –
4.1.2. Scope and purposeApart from the conceptual data model, its documentation seems
to provide a clear scope of its purpose. Furthermore, a detailedspecification document of the Rural-eGov Observatory system pro-vides a number of use cases that helped in eliciting some additionalrequirements for the development of the AP. The only limitationis that the introduction of the use cases is not in the documentthat describes the AP itself—this is a shortcoming that could beaddressed by the developers of the schema.
4.1.3. Selection of elementsThe ReGov LOM AP contains only elements from the IEEE LOM
Standard. In specific, from a total of 77 elements, ReGov LOM uses65 elements of LOM. Based on this selection, ReGov LOM can beconsidered as a subset AP of IEEE LOM.
4.1.4. Allowed modifications from base schemaContrary to the IEEE LOM Standard where the use of all of its
elements is optional, Regov LOM Application Profile defines 39mandatory elements, such as 1:General and 5:Educational. Addition-ally ReGov LOM reduces the size of some elements to one, contraryto LOM that defines their size as multiple, e.g. 1.1:General.Identifierand 1.4:General.Description.
On the other hand the value space of some elements is restrictedto a subset of the values that are provided by the IEEE LOM forthe given elements, e.g. value space of 2.3.1:LifeCycle.Contribute.Roleelement is restricted to “author”, “publisher”, “unknown”, “val-
idator”, “editor”, “subject matter expert”, “annotator”. Additionallyother elements’ value spaces are defined as references to anotherstandard, e.g. value space of 1.5:General.Keyword is defined fromthe AGRIS Subject Categories (http://www.fao.org/scripts/agris/c-categ.htm) of FAO. For this reason the value space of ReGov LOM
N. Manouselis et al. / Computers and Electronics in Agriculture 70 (2010) 302–320 307
f the
att
4
1TEcCoad
4
htitwtal
tr
Fig. 1. The conceptual schema o
pplication profile can be considered as multi-source. Overall, allhese modifications conform to the CEN/ISSS guidelines regardinghe modification of the value spaces.
.1.5. Non-allowed modifications from base schemaReGov LOM has changed the datatype of the elements
.5:General.Keyword, 1.6:General.Coverage, 9.2.2.1:Classification.axonPath.Taxon.Id and 9.2.2.1:Classification.TaxonPath.Taxon.ntry from ‘LangString’ to ‘Vocabulary’. This is a non-allowedhange: the modification of the datatypes is not conformant to theWA 15555 guidelines. It could be suggested to the developersf the AP to keep the datatype as ‘Langstring’, and to provide asrecommendation to the implementers that they allow only the
esired text values in the metadata input system interface.
.2. FAO Ag-LR
Capacity and institution building is a core function of FAO. FAOas recently launched the Capacity and Institution Building Portal1
o provide structured access to information on FAO’s capacity andnstitution building services and learning resources. To ensure thathe Portal can be searched by users and to enable interoperabilityith other recognized educational repositories, the FAO Agricul-
ural Learning Resources (Ag-LR) AP was created conforming tovailable and commonly used standards, to describe agricultural
earning resources (Stuempel et al., 2007).The Ag-LR AP of FAO was published in September 2007 and aimso serve as an international reference for designing and developingepositories of agricultural learning resources. Its schema is based
1 http://www.fao.org/capacitybuilding/.
ReGov LOM application profile.
mainly on the Dublin Core Metadata Element Set (DCMES) andthe Agricultural Metadata Element Set (AgMES), with additionalelements taken from the IEEE LOM Standard.
4.2.1. DocumentationThe conceptual schema of the FAO Ag-LR application profile is
illustrated in Fig. 2. XML instances of the FAO Ag-LR’s metadata arealso provided from the portal. Although an XML technical bind-ing has been developed for internal use, it is not made public forother potential implementers. The binding claims conformance tothe DCMES base schema.
4.2.2. Scope and purposeWithin the FAO Ag-LR AP documentation, clear statements about
its purpose are provided, along with a set of basic requirements thataims to cover, as well as a set of examples of its use within FAO’sportal. On the other hand additional requirements regarding the useof FAO Ag-LR’s elements could be presented through exemplary usecases.
4.2.3. Selection of elementsFAO Ag-LR AP is made up of 22 elements. As stated above
it is derived by combining elements from DCMES and AgMES.Additional elements from the IEEE LOM standard have also beenselected, mainly to capture the educational characteristics of theagricultural learning resources. Since its elements are selectedfrom more than one standard, the FAO Ag-LR application profile
can be considered as a multi-source application profile. The samealso applies for its value space, since its values come from vari-ous sources (such as the LOM or FAO namespaces). The process ofselecting the elements and value spaces of the FAO Ag-LR AP isconformant to the CWA 15555 guidelines.
308 N. Manouselis et al. / Computers and Electronics in Agriculture 70 (2010) 302–320
of the
4
mIr
fav4Ca
4
a
4
(riCwcudv
LCrdIattdp
S
Fig. 2. The conceptual schema
.2.4. Allowed modifications from base schemaThe AP defines 9 mandatory elements, restricting the require-
ent of the base schemas that define all their elements as optional.n addition, it does not break any super-class/sub-class semanticelations.
Modifications to the value spaces of the elements are also con-ormant to the CWA 15555 guidelines, since they were defined asreference to existing specifications or as a subset of the defined
alue space of the base schema. For example the values of element.Subject/Categories are defined from FAO’s Technical Knowledgelassification Scheme2 while the values of 19.Intended End User Rolere restricted to “teacher”, “learner”, and “manager”.
.2.5. Non-allowed modifications from base schemaThe analysis has shown that there have not been any non-
llowed modifications of the base schemas in FAO’s Ag-LR AP.
.3. CG LOM core
The Consultative Group for International Agricultural ResearchCGIAR) is a strategic alliance of countries, international andegional organizations, and private foundations supporting 15nternational agricultural research centers in developing countries.GIAR has initiated the Online Learning Resources (OLR) projecthich addresses the need of an international teaching and learning
ommunity of practice, interested in tropical agriculture and nat-ral resources management research and development, to easilyiscover and retrieve relevant learning resources produced by thearious centers in collaboration with their national partners.
In the context of the OLR project, CGIAR published the CGOM Core application profile to enable the description of theGIAR learning resources for populating the CGIAR learning objectepository.3 CG LOM Core is developed upon the IEEE LOM stan-ard and was published on November 2005 (Zschocke et al., 2005).
t generally tries to cover the needs of an international teaching
nd learning community of practice, interested in tropical agricul-ure and natural resources management research and development,o easily discover and retrieve relevant learning resources pro-uced by the various centers in collaboration with their nationalartners.2 http://www.fao.org/aims/ag classifschemes.jsp?myLangTerms=EN&mychemeID=5.3 http://learning.cgiar.org.
FAO Ag-LR application profile.
4.3.1. DocumentationWithin its documentation, the conceptual data model of the CG
LOM Core application profile is being thoroughly presented. On theother hand no technical binding is publicly available, although thedevelopers of the schema claim technical conformance to the baseschema.
4.3.2. Scope and purposeA clear purpose statement of the CG LOM Core is being pro-
vided within its documentation. On the other hand, no use casesare included to capture/illustrate the specific needs of the targetedcommunity.
4.3.3. Selection of elementsCG LOM Core application profile’s elements are directly derived
from the IEEE LOM Standard. In specific, from a total of 77 ele-ments, CG LOM Core uses 74 elements of IEEE LOM standard. Forthis reason, CG LOM Core can be considered as a subset AP of IEEELOM.
4.3.4. Allowed modifications from base schemaContrary to the IEEE LOM Standard where the use of all of its
elements is optional, CG LOM Core defines 36 mandatory elements.On the other hand the value space of the application profile can beoverall considered as multi-source. On one hand, the majority of itselements use the value space as it has been defined by the IEEE LOMStandard. On the other hand, one element, namely 7:Relation.Kindextends the predefined value space with additional two elementsand another one, 9:Classification.TaxonPath.Taxon uses the values ofAGRIS Subject Categories as its value space.
4.3.5. Non-allowed modifications from base schemaThe analysis showed that while the obligation status of
2.3.1:LifeCycle.Contribute.Role, 2.3.1:Life Cycle.Contribute.Entity and2.3.1:LifeCycle.Contribute.Date has been modified to mandatorytheir parent element 2.3:LifeCycle.Contribute remained its status asoptional. This is not conformant to the CWA 15555 guidelines andit should be changed to mandatory.
Moreover, the size of the elements 5.9:Educational.
TypicalLearningTime and 5.10:Educational.Description has beenmodified from one to multiple. Again, this is a modification that isnot conforming to the CWA 15555 guidelines.In addition, the modification of datatypes from ‘Langstring’ to‘Vocabulary’ for elements 5.9:Educational.TypicalLearningTime and
N. Manouselis et al. / Computers and Electronics in Agriculture 70 (2010) 302–320 309
of the
5cami
viivi
4
bfcLogRpt
4
bot
4
t
4
S
Fig. 3. The conceptual schema
.10:Educational.Description is also a non-allowed modification. Itould be suggested to the developers of the AP to keep the datatypes ‘Langstring’, and to provide as a recommendation to the imple-enters that they allow only the desired text values in the metadata
nput system interface.Finally, as stated above the addition of two more values in the
alue space of element 7:Relation.Kind is again a non-allowed mod-fication since they do not derive from a standard value space. Its suggested that the developers of the schema make their newalue spaces available in a public namespace, in order for othermplementations to be able to find, reference, and use them.
.4. BIOAGRO LOM
BIOAGRO LOM is an IEEE LOM-based Application Profile that haseen developed in the context of the Bio@gro eContent project, to
acilitate the annotation/description of learning resources that areollected and described in the Bio@gro Web Portal.4 The BIOAGROOM AP has been particularly developed to support the descriptionf learning resources on the topic of organic agriculture. Multilin-ual descriptions in four languages (i.e. English, German, Greek andomanian) are stored in the portal. The application profile has beenublished on 20 December 2005 by the Informatics Laboratory ofhe Agricultural University of Athens (Bio@gro, 2005).
.4.1. DocumentationWithin its documentation the conceptual data model of the AP is
een thoroughly presented. The tree view of the conceptual schemaf the BIOAGRO LOM AP is provided in Fig. 3. On the other hand noechnical binding of the application profile is being provided.
.4.2. Scope and purposeThe BIOAGRO LOM AP documentation seems to describe clearly
he scope of its purpose. However, it does not include any use cases.
.4.3. Selection of elementsThe BIOAGRO LOM AP contains only elements from the IEEE LOM
tandard. In specific, from a total of 77 elements, BIOAGRO LOM uses
4 http://www.bioagro.gr.
BIOAGRO application profile.
35 elements of LOM. For this reason, BIOAGRO LOM can be consideredas a subset AP of IEEE LOM.
On the other hand, the value set of this AP can be consid-ered as ad hoc since the value space of several elements (such as1.5:General.Keyword, 5.5:Educational.IntendedEndUserRole) has beencreated to serve the specific requirements of the project. Howeverthe rest of the elements (with two exceptions that are discussedbelow) follow the complete value space as it has been defined bythe IEEE LOM.
4.4.4. Allowed modifications from base schemaBIOAGRO LOM defines 16 mandatory elements. Additionally it
reduces the size of some elements to one, contrary to IEEE LOMthat defines their size as multiple, e.g. 1.4:General.Description and2.3:Life Cycle.Contribute.
4.4.5. Non-allowed modifications from base schemaBIOAGRO LOM has made non-allowed data type modi-
fications from ‘Langstring’ to ‘Characterstring’ in elements1.2:General.Title, 1.4:General.Description, 4.6:Technical.Other-PlatformRequirements, 5.6:Educational.Typical Age Range and6:Rights.Description). It would be an option for the developers tomake sure that when they extract content and/or exchange meta-data with other repositories, they make sure that the stored valuesfor these elements are transformed into ‘Langstring’ datatypes.
This AP also changed the datatype of other elements from‘Langstring’ to ‘Vocabulary’. It could be suggested to the develop-ers of the AP to keep the datatype as ‘Langstring’, and to provideas a recommendation to the implementers that they allow only thedesired text values in the metadata input system interface.
Additionally, the value space of elements 3.2.1:Meta-Metadata.Contribute.Role and 5.2:Educational.LearningResourceType(as been defined by IEEE LOM) has been extended with additionalvalues to meet the needs of the specific project. However thismodification does not conform to the CWA 15555 guidelines sincethese values are not referenced from any other known standardor specification nor they were published in a public namespace.
It is suggested that the developers of the schema make their newvalue spaces available in a public namespace, in order for otherimplementations to be able to find, reference, and use them.Finally, the BIOAGRO LOM AP has modified the cardinalityof 18 elements of the base schema. However the cardinality
3 lectron
mgnaabib
4
aloeAllb
borhts
4
AN
4
gdw
4
hcac
cduteu
4
iIhhSt
formant to the simple DCMES. Additionally, some DC qualifiers havealso been included (e.g. elements Spatial and Temporal as refine-ments of DC element Coverage, and element IsPartOf as refinement
10 N. Manouselis et al. / Computers and E
odification of 3 elements was not in line with the CWA 15555uidelines. In specific, contrary to the base schema, the cardi-ality of elements 2.3.1:LifeCycle.Contribute.Role, 4.2:Technical.Sizend 4.6:Technical.OtherPlatformRequirements has been extended tosize larger than one. This is a non-allowed modification that has toe carefully considered when extracting content and/or exchang-
ng metadata with other repositories. Inevitably, information wille lost during such a transformation/mapping.
.5. BEN
The Biosci Education Network5 (BEN) Collaborative establishedportal site and is developing and maintaining digital library col-
ections of biological sciences teaching and learning resources. Partf the development of the portal and digital library collectionsntailed the establishment of a metadata specification (the BENP) that all partners agreed to adhere to for describing their col-
ection’s material. BEN AP aims to support the cataloging of BENearning resources and user discovery (searching and browsing) ofiology teaching and learning resources.
BEN application profile was published on April 2003 and it isased on a previous version of IEEE LOM Standard: more precisely,n April’s 2001 version 6.1 a direct precedent version of the cur-ent standard. Their difference lies in the naming of some elementsowever their semantics remain the same. For example, the Iden-
ifier element is called CatalogEntry but they both have the sameemantics.6
.5.1. DocumentationThe documentation describes a clear scope and purpose of the
P and its conceptual data model is being thoroughly presented.o technical binding of the conceptual data model is provided.
.5.2. Scope and purposeAlthough the document states that ad hoc specialized working
roups were teamed up to discuss and conclude on the conceptualata model, details on how user requirements were collected alongith some indicative use cases could also be provided.
.5.3. Selection of elementsAll available elements from the IEEE LOM v6.1 metadata schema
ave been selected for the BEN AP, and for this reason it can beonsidered as a complete AP. However, it cannot be considered asconforming IEEE LOM AP, due to structural differences with the
urrent version of the standard.On the other hand, regarding the value space of the BEN appli-
ation profile, in some occasions ad hoc vocabularies had to beeveloped in order to cover the particular requirements of BEN’sser groups, e.g. 5.2:Educational.LearningResourceType, 5.6:Educa-ional.Context and 9.1:Classification.Purpose. For the rest of thelements the complete value space of IEEE LOM’s v6.1 is beingsed.
.5.4. Allowed modifications from base schemaContrary to the IEEE LOM v.6.1 schema where the use of all of
ts elements is optional, BEN AP defines 31 mandatory elements.n addition, some modifications to the value spaces of elements
ave taken place. For instance, the value space of 2.2:LifeCycle.Statusas been modified to the one used by the Digital Library for Earthystem Education (DLESE) metadata specification. In 2.3.1:Con-ribute.Role, BEN uses a subset of the LOM-proposed vocabulary.5 http://www.biosciednet.org/.6 Note: In IEEE LOM version 6.1, element 1.1:Identifier was not available for use.
ics in Agriculture 70 (2010) 302–320
4.5.5. Non-allowed modifications from base schemaAs noted above, in two elements, ad hoc value spaces have been
defined in order to meet the needs of the user community. Morespecifically, these have been 5.2:Educational.Learning and 5.6.Edu-cational.Context. We could not locate their value spaces in somepublic namespace. Therefore, it would be suggested to the devel-opers of the schema to make their own value spaces available in apublic namespace, in order for other implementations to be able tofind, reference, and use them.
4.6. SANREM CRSP
The Sustainable Agriculture and Natural Resource ManagementCollaborative Research Support Program (SANREM CRSP) is spon-sored by the U.S. Agency for International Development’s EconomicGrowth, Agriculture, and Trade Bureau (USAID/EGAT) and partic-ipating U.S. and host country institutions around the world. Theobjective of the SANREM CRSP is to support sustainable agricultureand natural resource management decision makers in developingcountries by providing access to appropriate data, knowledge, tools,and methods of analysis; and by enhancing their capacity to makebetter decisions to improve livelihoods and the sustainability ofnatural resources.
All SANREM programs and activities contribute to the onlineSANREM Knowledgebase (SKB)7 which it is intended to serve as acatalog of information resources specific to the SANREM project aswell as catalog and archive other resources and projects that relateto sustainable agriculture and natural resource management. The“resources” cataloged in the SKB (using DC-based metadata) areprimarily articles, papers, and reports but may include other digi-tal resources such as presentations, images, web pages, and othermaterials that can be referenced. The overall goal is to make theseresources readily available to facilitate the wide and effective dis-semination of information and to provide a structure for effectivesearch and retrieval of the resources. The SANREM AP is definedin the SANREM Knowledgebase Metadata Guide8 that has beenpublished in September 2007.
4.6.1. DocumentationWithin its documentation the conceptual data model of the
SANREM KB application profile is being presented. Additionally,conformance is claimed to the DCMES—however for sake of interop-erability, a more in depth analysis of the application profile couldbe presented following the CWA 14855 “Dublin Core ApplicationProfile Guidelines”.9 On the other hand, no technical binding ofthe conceptual data is provided, nor conformance is claimed to thetechnical binding of the base schema.
4.6.2. Scope and purposeA clear purpose statement of the application profile is being pro-
vided. However, no use cases are being provided for capturing therequirements of the targeted user groups.
4.6.3. Selection of elementsAs mentioned above, the SANREM KB application profile is con-
of DC element Relation. Additionally some ad hoc elements have
7 http://www.oired.vt.edu/sanremcrsp/menu information/SKB.php.8 http://www.oired.vt.edu/sanremcrsp/documents/SKB.metadata.guide.V4.Oct.
2007.pdf.9 ftp://cenftp1.cenorm.be/PUBLIC/CWAs/e-Europe/MMI-DC/cwa14855-00-
2003-Nov.pdf.
lectro
bcPpA
4
mDt
4
otblonr
4
gabwzos
o(i
4
tHibT
4
ad
4
hcfvit
4
d
panying documentation that described/specified the AP. In three(3) cases, there was no documentation. After communicating with
N. Manouselis et al. / Computers and E
een introduced for serving the specific needs of the SANREMommunity, such as the elements SANREMProductType, SANREM-rojectID and UploadResource. Overall, the SANREM KB applicationrofile can be considered as a mixture of a subset AP and an ad hocP of IEEE LOM.
.6.4. Allowed modifications from base schemaThe SANREM KB Application Profile defines 6 mandatory ele-
ents, namely Title, Description, Creator (author), CreationDate,escription and Type. No additional modifications have been made
o the value space of the base schema.
.6.5. Non-allowed modifications from base schemaA vocabulary set has been defined for describing the keywords
f a given resource. However this modification does not conform tohe CWA 15555 guidelines since this value space is not referencedy any other known standard or specification nor they were pub-
ished in a public namespace. It is suggested that the developersf the schema make their new value spaces available in a publicamespace, in order for other implementations to be able to find,eference, and use them.
.7. TragLor
The Turkish Agricultural Learning Objects Repository10 (TrA-Lor) is a Turkish initiative project, funded by the Scientificnd Technological Research Council of Turkey. It is coordinatedy Agricultural Faculty of Cukurova University. In collaborationith several educational, private and public sector organi-
ations, it aims to promote an infrastructure for learningbjects in agriculture, food, environment, forestry and veterinaryciences.
For the storage and the description of the associated learningbjects in the repository, the TrAgLor LOM AP has been developedCebeci et al., 2008). As with the BEN Application Profile, TrAgLors based on April’s 2001 version 6.1 of IEEE LOM Standard.
.7.1. DocumentationThe TrAgLor project team has published some research papers
hat present a detailed architectural analysis of the TrAgLor project.owever, there is no formal and sufficient documentation describ-
ng the TrAgLor LOM AP itself so far. On the other hand, the XMLinding of a TrAgLor LOM AP metadata instance is available fromrAgLor.
.7.2. Scope and purposeAlthough TrAgLor-relevant papers describe clearly the scope
nd purpose of the repository, there is no documentation availableescribing the scope and purpose of the metadata AP itself.
.7.3. Selection of elementsAll available elements from the IEEE LOM v6.1 metadata schema
ave been selected for the TrAgLor AP, and for this reason it can beonsidered as complete. However, it cannot be considered as a con-orming IEEE LOM AP, due to structural differences with the currentersion of the standard. Due to the lack of sufficient documentation,t is not possible to draw any conclusions about the value spaces ofhe elements.
.7.4. Allowed modifications from base schemaAgain, no solid conclusions can be drawn up, due to insufficient
ocumentation.
10 http://traglor.cu.edu.tr/.
nics in Agriculture 70 (2010) 302–320 311
4.7.5. Non-allowed modifications from base schemaIn this case as well, no solid conclusions can be made.
4.8. Intute: Health and Life Sciences
Intute11 is a free Web-enabled service aimed at students, teach-ers, and researchers in UK further education and higher education.Intute provides to online access to a large database of resourcesthat cover four main subjects: (a) Science and Technology, (b) Artsand Humanities, (c) Social Sciences and (d) Health and Life Sci-ences. One particular collection of the last subject is the one listingresources on Agriculture, Food and Forestry—formerly known asthe AgriFor service.12 In early June 2008 Intute provided 123,381records. Intute is not to be confused with a simple search engine,since subject experts continuously select and include resourcesdescribing them with appropriate metadata. Agriculture-relatedannotation also takes place, for instance according to the CAB The-saurus (http://www.cabi.org/DatabaseSearchTools.asp?PID=277).The Intute AP is based on DC. Due to insufficient documentation, afurther analysis of the AP was not possible.
4.9. EcoLearnIT
EcoLearnIT13 is a digital repository of reusable learning objectsthat manages and hosts various resources focused on soil, water andenvironmental sciences, and provides authoring tools to developlearning objects. It is an open-access system in which learningobjects are created, reviewed and used by an international com-munity of online learners, students, instructors and scientists.EcoLearnIT facilitates learning at all levels ranging from simple tocomplex knowledge encapsulated into different types of resources,targeting various learning audiences (ranging from post-graduate,graduate and undergraduate students, online learners, farmers,ecologists, consultants, agencies and others). EcoLearnIT is imple-mented in form of an online journal that provides full credit toauthors and co-authors (i.e. the learning object developer). Theresources stored in EcoLearnIT can be cited and referenced (hyper-linked) to be included in courses, training and extension materialand accessed by students, learners and instructors. The EcoLearniIT metadata AP is based on IEEE LOM. Due to insufficient documen-tation, a further analysis of the AP was not possible.
5. Outcomes
This section discusses the overall observations of the previousanalysis, and integrates the suggestions upon each AP into a set ofgeneric recommendations that could be useful for the designersand developers of metadata APs for AgLRs.
5.1. General observations
The analysis of the sample of studied APs has led to a numberof interesting observations, as far as the CWA 15555 guidelines areconcerned.
5.1.1. DocumentationThe majority of the nine (9) identified APs for AgLRs had accom-
the operators of the repositories, we have been informed that forTrAgLor and EcoLearnIT, they intended to prepare and publish
11 http://www.intute.ac.uk/.12 http://www.intute.ac.uk/healthandlifesciences/agriculture/.13 http://EcoLearnIT.ifas.ufl.edu.
3 lectron
scde
(
•
•
•
5
sinF
5
c
•
•
•
sBCL
fw
ut
•
•
•
a
5
o
•
•
•••
12 N. Manouselis et al. / Computers and E
uch documentation in the near future. As far as Intute is con-erned, there was no plan for the preparation of a specific documentescribing the metadata AP, since it is based on the more genericPrints metadata schema (which is a DC AP).
For the seven (7) APs of the sample that have been analysedincluding TrAgLor to the extend this was possible):
All of them had a description or specification of their conceptualmodel, as requested by CWA 15555;Only two (2) of them (i.e. FAO’s Ag-LR, TrAgLor LOM) providepublic access to a technical binding (in both cases in XML).Three (3) of them claim technical conformance to their baseschema. That is, FAO’s Ag-LR to DC; CG LOM Core and TrAgLorLOM to IEEE LOM.
.1.2. Scope and purposeIn all seven (7) documented APs there is a description of the
cope and purpose, as suggested by CWA 15555. On the other hand,n only two (2) is there a set of use cases further elaborating theeeds and requirements of the user community (i.e. ReGov LOM,AO’s Ag-LR).
.1.3. Number of elementsAs far as the selected elements from the base schemas are con-
erned, the analysed APs can be classified as (Najjar et al., 2004):
Complete APs, three (3) of them. These are SANREM KB, BEN LOMand TrAgLor LOM.Subset APs, three (3) of them. These are ReGov LOM, CG LOM Core,and BIOAGRO LOM.Multi-source APs, one (1) of them. This is FAO’s Ag-LR one.
It was interesting to observe that in five (5) of the APs, the con-idered elements where either all the LOM ones (i.e. SANREM KB,EN, TrAgLor) or the majority of them (i.e. ReGov LOM, CG LOMore). These included more than 65 elements (out of the 77 thatOM has).
On the other hand, the other two (2) APs proposed the use ofewer elements. BIOAGRO LOM includes thirty-five (35) elements,
hereas FAO’s Ag-LR only twenty-two (22).Nevertheless, these numbers are indicative since the mandated
se of elements is not related to the potential number of elementshat an AP includes. In the examined sample:
Three (3) APs require more than thirty (30) mandatory elements:ReGov LOM (39), CG LOM Core (36), BEN LOM (31).One (1) AP requires more than fifteen (15) mandatory elements:BIOAGRO LOM (16).Two (2) APs require less than ten (10) mandatory elements: FAOAg-LR (9), SANREM KB (6).
For TrAgLor, no information about mandatory elements isvailable.
.1.4. Allowed modifications from base schemaThe allowed modifications from the base schema that have been
bserved most often are:
The mandatory selection of non-mandatory elements: in six (6)
APs.The change in the definition of the size and smallest permittedmaximum: in two (2) APs.The change in the obligation of data elements: in six (6) APs.The allowed modifications of value spaces: in five (5) APs.Other modifications: in one (1) AP.ics in Agriculture 70 (2010) 302–320
5.1.5. Non-allowed modifications from base schemaThe non-allowed modifications from the base schema that have
been observed most often are:
• The extension of the cardinality of elements: in one (1) AP.• The addition of new items in controlled vocabulary lists: in two
(2) APs.• The change in the mandatory status of an element: in one (1) AP.• The use of a base schema version other than the latest, stable one:
in two (2) APs.
Apart from those, the rest of non-allowed modifications thatCWA 15555 mentions have not been noted. That is:
• The change in the location of a data element.• The creation of new elements that mimic the semantics of existing
ones.• The change in the meaning of existing elements.• The change in the name of existing elements.• The extension of a base schema in other than specified points.
5.2. Mappings
Another outcome of the analysis was the elaboration of map-pings among the various application profiles. We chose to depictthose mappings in relation to the base schema that the applicationprofile conforms to. That is, Table 3 presents the mappings of theIEEE LOM-based schemas to IEEE LOM and between them. Table 4presents the mapping of the DC-based schema (SANREM KB AP) toDC. And Table 5 provides a mapping of the elements of the multi-source FAO AgLR AP with its respective metadata element sets,namely the DC, IEEE LOM, and the Agricultural Metadata ElementSet (AgMES, http://www.fao.org/aims/agmes elements.jsp) whichis a DC application profile for agricultural information resources.
5.3. Usage of elements
Based on the analysis of which elements are more often used inthe examined sample, we have concluded to some initial obser-vations about the elements that seem to be more popular inagricultural APs. More specifically, to examine the occurrence ofelements in the sample of APs for which we had enough documen-tation, a number of steps have been followed:
• The set of IEEE LOM elements are used are the reference ones,and all the elements that the examined APs used are mappedupon them.
• The number of times that a LOM element appears (as such in LOMAPs, or an equivalent one in other APs) has been counted. This hasbeen denoted as the metric COUNT.
• Since the appearance of an element does not give some indicationof its importance for the AP designers (e.g. it may be a mandatoryor an optional one), we have also counted whether an element hasbeen defined as mandatory, recommended or optional in an AP. Togive an indicative measure of this, we have defined another metricdenoted as WEIGHTED: this weights the occurrence of an elementdepending on if it is mandatory (where it is weighted with 1.5),recommended (where it is weighted with 1), or optional (whereit is weighted with 0.5). If the AP does not define a particularobligation status, then its occurrence is weighted with 1.
Fig. 4 illustrates the results of this analysis. For all IEEE LOM ele-ments, it shows how many times they appear in the studied sampleof APs (metric COUNT), as well as the weighted measure of theiroccurrence (metric WEIGHTED).
N.M
anouseliset
al./Computers
andElectronics
inA
griculture70 (2010) 302–320
313Table 3Mappings of LOM-based agricultural APs to LOM.
IEEE LOM ReGov LOM BIOAGRO LOM CG LOM Core BEN TrAgLOR
1. General 1 General M 1 General M 1. General M 1. General M 1. General1.1 Identifier 1.1 Identifier M Identifier M 1.1 Identifier M 1.3 Catalog Entry O 1.3 Catalog Entry1.1.1 Catalog 1.1.1 Catalog M 1.1.1 Catalog M 1.3.1 Catalog O 1.3.1 Catalog1.1.2 Entry 1.1.2 Entry M Entry M 1.1.2 Entry M 1.3.2 Entry O 1.3.2 Entry1.2 Title 1.2 Title M Title M 1.2 Title M 1.2 Title M 1.2 Title1.3 Language 1.3 Language M Language M 1.3 Language M 1.4 Language M 1.4 Language1.4 Description 1.4 Description M Description M 1.4 Description M 1.5 Description M 1.5 Description1.5 Keyword 1.5 Keyword M Keyword M 1.5 Keyword M 1.6 Keyword O 1.6 Keyword1.6 Coverage 1.6 Coverage M Coverage M 1.6 Coverage M 1.7 Coverage O 1.7 Coverage1.7 Structure 1.7 Structure O 1.7 Structure M 1.8 Structure O 1.8 Structure1.8 Aggregation Level 1.8 Aggregation Level O Aggregation Level M 1.8 Aggregation Level M 1.9 Aggregation Level O 1.9 Aggregation Level
2. Life Cycle 2. Life Cycle M 2. Life Cycle M 2. Life Cycle M 2 Life Cycle M 2. Life Cycle2.1 Version 2.1 Version O 2.1 Version O 2.1 Version O 2.1 Version2.2 Status 2.2 Status O 2.2 Status O 2.2 Status M 2.2 Status2.3 Contribute 2.3 Contribute O Contribute M 2.3 Contribute O 2.3 Contribute M 2.3 Contribute2.3.1 Role 2.3.1 Role M Role O 2.3.1 Role M 2.3.1 Role M 2.3.1 Role2.3.2 Entity 2.3.2 Entity M Entity M 2.3.2 Entity M 2.3.2 Entity M 2.3.2 Entity2.3.3 Date 2.3.3 Date M 2.3.3 Date M 2.3.3 Date M 2.3.3 Date
3. Meta-Metadata 3. Meta-Metadata M 3. Meta-Metadata O 3. Meta-Metadata M 3 Meta-metadata M 3. Meta-Metadata3.1 Identifier 3.1 Identifier M 3.1 Identifier M 3.2 Catalog Entry 3.2 Catalog Entry3.1.1 Catalog 3.1.1 Catalog M 3.1.1 Catalog M 3.2.1 Catalog 3.2.1 Catalog3.1.2 Entry 3.1.2 Entry M 3.1.2 Entry M 3.2.2 Entry 3.2.2 Entry3.2 Contribute 3.2 Contribute M Contribute O 3.2 Contribute M 3.3 Contribute M 3.3 Contribute3.2.1 Role 3.2.1 Role M Role O 3.2.1 Role M 3.3.1 Role M 3.3.1 Role3.2.2 Entity 3.2.2 Entity M Entity O 3.2.2 Entity M 3.3.2 Entity M 3.3.2 Entity3.2.3 Date 3.2.3 Date M Date O 3.2.3 Date M 3.3.3 Date M 3.3.3 Date3.3 Metadata Schema 3.3 Metadata Schema M Metadata Schema O 3.3 Metadata Schema M 3.4 Metadata Scheme M 3.4 Metadata Schema3.4 Language 3.4 Language M Language O 3.4 Language M 3.5 Language M 3.5 Language
4. Technical 4. Technical O 4. Technical M 4. Technical M 4. Technical M 4. Technical4.1 Format 4.1 Format O Format M 4.1 Format O 4.1 Format M 4.1 Format4.2 Size 4.2 Size O Size O 4.2 Size O 4.2 Size O 4.2 Size4.3 Location 4.3 Location O Location M 4.3 Location O 4.3 Location M 4.3 Location4.4 Requirement – – 4.4 Requirement O 4.4 Requirements O 4.4 Requirement4.4.1 OrComposite – – 4.4.1 OrComposite O4.4.1.1 Type – 4.4.1.1 Type O 4.4.1 Type O 4.4.1 Type4.4.1.2 Name – 4.4.1.2 Name O 4.4.2 Name O 4.4.2 Name4.4.1.3 Minimum Version – 4.4.1.3 Minimum Version O 4.4.3 Minimum Version O 4.4.3 Minimum Version4.4.1.4 Maximum Version – O 4.4.4 Maximum Version O 4.4.4 Maximum Version4.5 Installation Remarks – – 4.5 Installation Remarks O 4.5 Installation Remarks O 4.5 Installation Remarks4.6 Other Platform
Requirements4.6 Other PlatformRequirements
O Other PlatformRequirements
O 4.6 Other PlatformRequirements
O 4.6 Other PlatformRequirements
O 4.6 Other PlatformRequirements
4.7 Duration 4.7 Duration O Duration O 4.7 Duration O 4.7 Duration O 4.7 Duration
5. Educational 5. Educational M 5. Educational M 5. Educational M 5. Educational M 5. Educational5.1 Interactivity Type 5.1 Interactivity Type M 5.1 Interactivity Type M 5.1 Interactivity Type O 5.1 Interactivity Type5.2 Learning Resource Type 5.2 Learning Resource Type M Learning Resource Type M 5.2 Learning Resource Type M 5.2 Learning Resource Type M 5.2 Learning Resource Type5.3 Interactivity Level 5.3 Interactivity Level M 5.3 Interactivity Level M 5.3 Interactivity Level O 5.3 Interactivity Level5.4 Semantic Density – – 5.4 Semantic Density O 5.4 Semantic Density O 5.4 Semantic Density5.5 Intended End User Role 5.5 Intended End User Role M Intended End User Role M 5.5 Intended End User Role M 5.5 Intended End User Role M 5.5 Intended End User Role5.6 Context 5.6 Context M Context O 5.6 Context M 5.6 Context M 5.6 Context5.7 Typical Age Range 5.7 Typical Age Range O Typical Age Range O 5.7 Typical Age Range O 5.7 Typical Age Range O 5.7 Typical Age Range5.8 Difficulty 5.8 Difficulty O 5.8 Difficulty O 5.8 Difficulty O 5.8 Difficulty5.9 Typical Learning Time 5.9 Typical Learning Time O 5.9 Typical Learning Time O 5.9 Typical Learning Time O 5.9 Typical Learning Time5.10 Description – – 5.10 Description O 5.10 Description O 5.10 Language5.11 Language 5.11 Language O 5.11 Language O 5.11 Language M 5.11Description
314 N. Manouselis et al. / Computers and Electronics in Agriculture 70 (2010) 302–320Ta
ble
3(C
onti
nued
)
IEEE
LOM
ReG
ovLO
MB
IOA
GR
OLO
MC
GLO
MC
ore
BEN
TrA
gLO
R
6.R
igh
ts6.
Rig
hts
M6.
Rig
hts
M6.
Rig
hts
M6.
Rig
hts
M6.
Rig
hts
6.1
Co
st6.
1C
ost
MC
ost
M6.
1C
ost
M6.
1C
ost
M6.
1C
ost
6.2
Co
pyri
ght
and
Oth
erR
estr
icti
on
s6.
2C
opy
righ
tan
dO
ther
Res
tric
tio
ns
MC
opy
righ
tsan
dO
ther
Res
tric
tio
ns
M6.
2C
opy
righ
tan
dO
ther
Res
tric
tio
ns
M6.
2C
opy
righ
tan
dO
ther
Res
tric
tio
ns
M6.
2C
opy
righ
tan
dO
ther
Res
tric
tio
ns
6.3
Des
crip
tio
n6.
3D
escr
ipti
on
OD
escr
ipti
on
O6.
3D
escr
ipti
on
O6.
3D
escr
ipti
on
M6.
3D
escr
ipti
on
7.R
elat
ion
7.R
elat
ion
O7.
Rel
atio
nO
7.R
elat
ion
O7.
Rel
atio
n7.
1K
ind
7.1
Kin
dO
7.1
Kin
dO
7.1
Kin
dO
7.1
Kin
d7.
2R
eso
urc
e7.
2R
eso
urc
eO
7.2
Res
ou
rce
O7.
2R
eso
urc
eO
7.2
Res
ou
rce
7.2.
1Id
enti
fier
7.2.
1Id
enti
fier
O7.
2.1
Iden
tifi
erO
7.2.
3C
atal
ogEn
try
O7.
2.3
Cat
alog
Entr
y7.
2.1.
1C
atal
og7.
2.1.
1C
atal
ogO
7.2.
1.1
Cat
alog
O7.
2.3.
1C
atal
ogO
7.2.
3.1
Cat
alog
7.2.
1.2
Entr
y7.
2.1.
2En
try
O7.
2.1.
2En
try
O7.
2.3.
2En
try
O7.
2.3.
2En
try
7.2.
2D
escr
ipti
on–
–7.
2.2
Des
crip
tion
O7.
2.2
Des
crip
tion
O7.
2.2
Des
crip
tion
8.A
nn
otat
ion
8.A
nn
otat
ion
O8.
An
not
atio
nO
8.A
nn
otat
ion
O8.
An
not
atio
n8.
1E
nti
ty8.
1E
nti
tyO
8.1
En
tity
O8.
1P
erso
nO
8.1
Per
son
8.2
Dat
e8.
2D
ate
O8.
2D
ate
O8.
2D
ate
O8.
2D
ate
8.3
Des
crip
tio
n8.
3D
escr
ipti
on
O8.
3D
escr
ipti
on
O8.
3D
escr
ipti
on
O8.
3D
escr
ipti
on
9.C
lass
ifica
tio
n9.
Cla
ssifi
cati
on
M9.
Cla
ssifi
cati
on
M9.
Cla
ssifi
cati
on
M9.
Cla
ssifi
cati
on
9.1
Pu
rpo
se9.
1P
urp
ose
M9.
1P
urp
ose
M9.
1P
urp
ose
M9.
1P
urp
ose
9.2
Taxo
nP
ath
9.2
Taxo
nP
ath
M9.
2Ta
xon
Pat
hM
9.2
Taxo
nP
ath
M9.
2Ta
xon
Pat
h9.
2.1
Sou
rce
9.2.
1So
urc
eM
9.2.
1So
urc
eM
9.2.
1So
urc
eM
9.2.
1So
urc
e9.
2.2
Taxo
n9.
2.2
Taxo
nM
9.2.
2Ta
xon
M9.
2.2
Taxo
nM
9.2.
2Ta
xon
9.2.
2.1
Id9.
2.2.
1Id
M9.
2.2.
1Id
M9.
2.2.
1Id
M9.
2.2.
1Id
9.2.
2.2
Entr
y9.
2.2.
2En
try
M9.
2.2.
2En
try
M9.
2.2.
2En
try
M9.
2.2.
2En
try
9.3
Des
crip
tio
n–
–9.
3D
escr
ipti
on
O9.
3D
escr
ipti
on
O9.
3D
escr
ipti
on
9.4
Key
wo
rd–
–9.
4K
eyw
ord
O9.
4K
eyw
ord
sM
9.4
Key
wo
rds
Table 4Mapping of SANREM KB AP to DC.
SANREM KB AP Dublin Core MES
Title dc:titleAlternate title dcterms:alternativeCreator (author) dc:creatorContributor dc:contributorContact informationLandscape systemPES resource/projectRestricted keywords dc:subjectUnrestricted keywords dc:subjectDescription dc:descriptionPublisher dc:publisherBibliographic citationIs part of dcterms:isPartOfCreation date dcterms:createdType dc:typeSANREM product typeFormat dc:formatIdentifier dc:identifierURL dc:identifierLanguage dc:languageSpatial dcterms:spatialTemporal dcterms:temporalRights dc:rights
Upload resourceSANREM project IDSome interesting observations can be made from this analysis:
• Most of the APs are using some element to store an identifier ofthe resource. In some cases, this is only a URL (in other cases, aformal catalog system can also been used).
• As far as the rest of the general characteristics of the resource areconcerned, the following information is usually captured:◦ Title;◦ Language;◦ Description;◦ Keyword (free text or restricted);◦ Coverage (geographical/spatial or temporal).
• As far as the life cycle of the resource is concerned, the followinginformation is usually captured:◦ Role of the entities that have contributed to the resource;◦ Information about these entities;◦ Date of contribution/production/publication.
• As far as the technical characteristics of the resource are con-cerned, the following information is usually captured:◦ Technical format;◦ Technical location (such as URL), when the Identifier element
is not used for this purpose;◦ Size;◦ Some technical requirements for its viewing/execution.
• As far as the educational characteristics of the resource are con-cerned, the following information is usually captured:◦ Type of the learning resource;◦ Intended end user role;◦ Educational context/level.
• As far as the copyrights of the resource are concerned, the follow-ing information is usually captured:◦ Cost;
◦ Copyrights and restrictions in use.• As far as the formal classification of the resource is concerned, thefollowing information is usually captured:◦ Purpose of classification;◦ The classification system used;◦ Terms used from the selected classification system.
N. Manouselis et al. / Computers and Electronics in Agriculture 70 (2010) 302–320 315
Table 5Mapping of FAO’s multi-source AP to its respective base schemas.
FAO AgLR AP Dublin Core MES IEEE LOM AGMES
Title dc:titleSupplement title dc:titleCreator dc:creatorSubject/FAO categories ags:subjectClassificationSubject/keywords ags:subjectThesaurusAbstract dcterms:abstractNotes ags:DescriptionNotesPublisher dc:publisherDate dcterms:issuedType dc:typeFormat dc:formatSize 4.2.Technical.SizeIdentifier dc:identifierLanguage dc:languageRelation dc:relationRelation: collection dcterms:hasPart dcterms:isPartOfRelation: language version ags:relationIsTranslationOf
ags:relationHasTranslationCoverage dcterms:spatialRegionCountryRights ags:rightsStatementCost 6.1.Rights.CostI 5.5.EC 5.6.EI 5.3.ET 5.9.E
5
otA
ntended end user roleontext
nteractivity levelypical learning time
.4. Suggestions to AP developers
Based on the analysis of the sample of APs and the overallbservations made above, we introduce the following sugges-ions/recommendations to the designers/developers of metadataPs for AgLRs:
Fig. 4. IEEE LOM elements’ usage in a
ducational.Intended End User Roleducational.Contextducational.Interactivity Levelducational.Typical Learning Time
1. Always provide supportive documentation describing the AP. Sup-
portive documentations offer and allow an overview for theselection and reference for detailed analysis within the adoptionphase.2. Include in documentation reference to the technical implementationof the AP and provide any relevant technical bindings. References
gricultural application profiles.
3 lectron
3
4
5
6
8
6
6
lbaAti(AtsCb
iLsbf
16 N. Manouselis et al. / Computers and E
to technical implementations and provided technical bindingsfacilitate the implementation and the technical interoperabilityand validity of the metadata instances.
. Include in documentation supportive use cases that help clarify itsscope, purpose and users. Use cases support the selection processduring the comparison of AP candidates and provide informationabout implementation potentials.
. Use the latest and more stable version of the base schema available.Different versions of metadata specifications and/or standardsoften have important differences that do not ensure backwardscompatibility. When starting an implementation project, it issuggested that AP designers/implementers chose the latest andmore stable version of the base schema that is publicly available.For instance, in one examined case, although the project was ini-tiated after the publication of the IEEE LOM standard in 2002, aprevious version of LOM has been used.
. When ad hoc or extended value spaces are used for some elements, itis required to make the new value spaces available in a public names-pace, in order for conformance to be maintained. Public availabilityis needed to ensure interoperability of future APs by allowingreferences to these published namespaces.
. Always use the datatype of the standard. Instead of substituting the‘Langstring’ datatype with the simpler ‘Characterstring’, it is sug-gested that simplicity is sought though appropriate interface design.For instance, when the type is changed from ‘Langstring’ to ‘Char-acterstring’, then implementers have to make sure that duringa transformation/mapping the stored values for these elementsare transformed into ‘Langstring’ datatypes in order to avoidinformation loss.
7. The non-allowed modification rules of CWA 15555 should be care-fully respected, because breaking them can lead to problems whentrying to export/exchange metadata. For instance, an extensionto the cardinality of an element can lead to loss of informationduring a transformation/mapping.
. The elements most often occurring as mandatory in the existing APsshould be considered for use also in other APs, to increase exchangeof information and interoperability. It is most probable that theinformation that is considered important in all other agricul-tural APs will also be important for a new one as well. To achieveinteroperability in metadata exchange, information about a char-acteristic that is stored in all other APs will have to be stored fora new AP as well.
. Discussion
.1. Related work
The work that has been carried out in this study is along theines of the examination of interoperability issues in LRs that haseen carried out by Najjar et al. (2004). Up to our knowledge, therere no similar studies that have taken place for the agriculturalPs, therefore this is a novel effort for the study and harmoniza-
ion of the implementation of metadata schemas in LRs. Somenitial steps to this direction have been made by Manouselis et al.2009), where two different agricultural APs have been compared.
more conceptual discussion about the issues and the difficul-ies that implementers of agricultural LRs face is presented at theummarizing report of the 1st Agricultural Learning Repositories E-onference (AgLR 2008, http://aglr.aua.gr/node/24) that was editedy Manouselis and Salokhe (2008).
As far as generic metadata APs for LRs are concerned, an interest-
ng study of the actual usage of the elements of LOM APs in existingRs has been made by Friesen et al. (2003). Another approach is thetudy of the technical conformance of the bindings of APs with theinding of the base schema, e.g. by testing it with a number of con-ormant software tools (Nadolski et al., 2006; Zervas and Sampson,ics in Agriculture 70 (2010) 302–320
2008). Nevertheless, such an analysis (based on the examination ofconformance of the metadata APs to the base schemas) has not – toour knowledge – been carried out in the past.
6.2. Benefits and limitations of this study
This study is the first that applies the CEN WS-LT methodologyfor assessing the implementation status of metadata applicationprofiles in a given domain. Thus, it provides a good case studyupon which similar studies can be conducted, and illustrates thetype of results that can be produced. The most important out-come is an indication of the harmonization steps that are requiredfor interoperable agricultural APs. Apart from complying with theconformance rules of the base schemas, the identification of theelements often used in most of the studied APs provides an indica-tion of the basic elements which all agricultural APs should include.If combined with an empirical study on how the elements are actu-ally used (populated) by the metadata annotators (Najjar, 2008),this work can provide valuable input towards a meta-model of theelements that most AgLRs use.
On the other hand, we cannot assume that these results canbe generalized in non-agricultural domains. The reason is that inagricultural education and training, learning technologies’ spec-ifications and standards have not yet been widely adopted. Fewinitiatives have reported implementing such standards, and in mostcases only the ones that describe learning resources by using IEEELOM or DC (Manouselis et al., 2009). Therefore, this might accountfor a high number of implementations that suffer from several inter-operability problems.
In addition, another limitation could be that the analysis carriedout in this study was based on the documentation at hand aboutthe APs. In several cases we had missing or incomplete informa-tion about the APs, which resulted in outcomes that were based onthe APs that had some documentation. These APs are usually themost developed ones, whereas the non-documented APs are theones, in most cases, are in need for additional feedback and sugges-tions. Nevertheless, having a more generic meta-model in place,together with a set of good examples (similar to some of the onescovered in this paper), can help AgLR implementers build bettersystems.
An additional limitation of this study is that it did not extendits analysis to the level of how agricultural knowledge organizationschemes (such as vocabularies or thesauri) are used in the sample.Several APs use their own vocabularies for classifying resourcesaround thematic areas, and it would be interesting to see whichones are the most prevalent ones (e.g. FAO’s AGROVOC thesaurus,http://www.fao.org/aims/agrovoccs.jsp).
Finally, the coverage of the analysis managed to include twoLRs from Asia, three from the US and four from Europe. The rea-son is that it was not possible to identify more AgLRs, althoughan open call for interest has been posted to relevant mailing listsand involved experts from more than 25 countries all over theworld (http://aglr.aua.gr). Nevertheless, this creates no bias to theproduced recommendations for AgLR metadata AP implementers,since they serve as guidelines to people developing AgLRs indepen-dently from the continent they are based at. It would be interestingand useful though to include more AgLRs in the sample as they arebeing developed and deployed.
6.3. Generalization
In general this study could show the current challenges for build-ing and adapting metadata APs with special emphasis on the field ofagricultural education and training. The benefits of guidelines likethe CWA 15555 are demonstrated to provide guidance and supportfor tailor-made and branch-specific AP. But the CWA 15555 is not
lectro
wi
attfhottl(whtc
t“scaesswi
7
us
N. Manouselis et al. / Computers and E
ell recognised until now and lacks more detailed explanations andnstructions.
The current main problems are: (i) the development of guidesnd rules for building conformant metadata APs with global accep-ance and (ii) the design and implementation of correspondingools to facilitate both the building of metadata APs and their con-ormance testing. Findings from other standardization areas mayelp to achieve these objectives: For quality standards a new classf decision-support management systems has been developed forhe adoption, implementation, and adaptation of quality standards:he Support Systems for Quality Management (QSS) with a threeevel approach for the tailor-made adaptation of quality standardsStracke and Hildebrandt, 2007). As pointed out, a general frame-ork for metadata APs is needed to introduce and ensure globalarmonization: It could benefit from a specific decision-supportool that enables users and organizations to build and test easilyonformant metadata APs.
First steps have been started in European standardization (c.f.he MLO standard in the standardization committee CEN/TC 353ICT for Learning, Education & Training”) as well in internationaltandardization (c.f. the MLR standard in the ISO standardizationommittee SC36 “IT for Learning, Education & Training”): That willlso facilitate the implementations of metadata APs for agriculturalducation and training in the future. The agreement on branch-pecific guides (like agricultural ontologies and schemas) couldupport in parallel the vision of the broad usage of metadata APsorldwide to enhance and optimize learning, education, and train-
ng by valuable and helpful standards.
. Conclusions
In this paper, we present results of a study that aimed at eval-ating existing implementations of metadata APs in AgLRs. Morepecifically, the paper assesses the current status of development
nics in Agriculture 70 (2010) 302–320 317
and implementation of metadata standards and specifications(such as IEEE LOM and DC), in the case of describing agriculturallearning resources. This study includes the selection and in depthanalysis of a sample of representative APs that have been imple-mented in repositories around the world. Then, it assesses thecompliance of the developed APs with their base schemas, andintroduces some initial recommendations regarding the betterdesign and implementation of such APs.
Our future work focuses on complementing this study with twoother testing activities. First, the conformance testing of technicalbindings, using a number of conformant software tools as test-beds.Second, the study of the actual usage of the metadata elements inthe targeted AgLRs, in order to identify the elements that the usersconceive as most important, or most relevant to their needs. Suchprocess can give feedback to the actual metadata AP implementa-tion phase, as well as to the definition of a reference meta-modelfor this domain. This will complement the work presented in thispaper, and could potentially benefit LR implementations in the agri-cultural sector as a whole.
Acknowledgments
The authors would like to thank the experts of the CEN WS LTfor their valuable comments and feedback. Part of the work pre-sented in this paper has been funded by the Food and AgricultureOrganization of the United Nations (FAO).
Appendix A.
In this annex, the tool used for the analysis of agricultural APs isincluded. The tool is based on the guidelines of CWA 15555 (2006)and Najjar et al. (2004). Apart from the following information, italso included a section where the analyst described the mapping ofthe studied AP to its base standard.
3 lectronics in Agriculture 70 (2010) 302–320
18 N. Manouselis et al. / Computers and E
lectronics in Agriculture 70 (2010) 302–320 319
R
B
B
B
B
C
tee for Standardization (CEN), May. (http://www.cen-isss-wslt.din.de/sixcmsupload/media/3050/CWA15966.pdf).
Dublin Core Metadata Element Set Version 1.1, 2004. Reference Descrip-
N. Manouselis et al. / Computers and E
eferences
alanskat, A., Vuorikari, R., 2000. Survey on School Educational Repositories (D 2.1).European Treasury Browser, European Schoolnet.
erge, Z.L., Leary, J., 2006. Trends and challenges of eLearning in national andinternational agricultural development. International Journal of Education andDevelopment using Information and Communication Technology 2 (2), 51–59.
io@gro, 2005. Metadata Models for Bio@gro Content Objects (BCOs) Description.Bio@gro Technical Report.
ueno-de-la-Fuente, G., Hernández-Pérez, T., Rodríguez-Mateos, D., Méndez-Rodríguez, E.M., Martín-Galán, B., 2009. Study on the use of metadata for digital
learning objects in institutional university repositories (MODERI). Catalogingand Classification, Special Issue on Metadata and Repositories, 262–285, 47(3/4).ebeci, Z., Y Erdogan, Z., Kara, M., 2008. TrAgLor: A LOM-Based Digital Learn-ing Objects Repository for Agriculture. Proc. of the 4th International ScientificConference on “eLearning and Software for Education” (eLSE’08), Bucharest,Romania, 17–18 April. (http://traglor.cu.edu.tr/objects/pdf/Traglor-else08.pdf).
CWA 14855, 2003. Dublin Core Application Profile Guidelines. European Com-mittee for Standardization (CEN), November 2003. (http://cenftp1.cenorm.be/PUBLIC/CWAs/e-Europe/MMI-DC/cwa14855-00-2003-Nov.pdf).
CWA 15555, 2006. Guidelines and support for building application pro-files in e-learning. European Committee for Standardization (CEN), June.(http://www.cen-isss-wslt.din.de/sixcms upload/media/3050/CWA15555.pdf).
CWA 15966, 2009. Guidelines and recommendations for building metadataapplication profiles for agricultural learning resources. European Commit-
tion, Retrieved 25 November 2008 from http://dublincore.org/documents/2004/12/20/dces/.
Duncan, C., 2002. Digital repositories: the ‘back-office of e-Learning or all e-Learning?’. In: Proc. of ALT-C 2002, Sunderland, 9–11 September.
Duval, E., et al., 2001. The ARIADNE knowledge pool system. Communications of theACM 44 (5), 73–78.
3 lectron
D
D
E
E
F
F
H
H
H
H
I
I
K
M
M
M
M
M
MN
N
N
Zervas, P., Sampson, D., 2008. A conformance testing approach with IEEE LOM and itsapplication to a science education application profile. In: Proc. of the 11th IASTED
20 N. Manouselis et al. / Computers and E
uval, E., Hodgins, W., Sutton, S., Weibel, S., 2002. Metadata principles and practi-calities. D-Lib Magazine 8 (3).
uval, E., Smith, N., Van Coillie, M., 2006. Application profiles for learning. In:Proceedings of the IEEE International Conference on Advanced Learning Tech-nologies (ICALT), pp. 242–246.
dReNe, 2008. Current state of educational repositories. Available fromhttp://edrene.org/results/index.html.
uzenat, J., 2002. An infrastructure for formally ensuring interoperability in a het-erogeneous semantic web. In: Cruz, et al. (Eds.), The Emerging Semantic Web,Selected Papers from the 1st Semantic Web Working Symposium, IOS Press.
riesen, N., 2001. What are educational objects? Interactive Learning Environments9 (3), 219–230.
riesen, N., Nirhamo, L., Knoppers, J., 2003. International Survey of Learn-ing Object Metadata Implementation for ISO/IEC. (http://mdlet.jtc1sc36.org/doc/SC36 WG4 N0029.pdf).
aughey, M., Muirhead, B., 2004. Evaluating Learning Objects for Schools. e-Journalof Instructional Science and Technology 8 (1), University of Southern Queens-land, Australia.
eery, R., 2002. Application profiles: interoperable friend or foe? In: TEL MilestoneConference, The European Library, pp. 34–39.
odgins, W.H., 2002. The future of learning objects. In: Proceedings of e-Technologiesin Engineering Education: Learning Outcomes Providing Future Possibilities, pp.76–82.
olden, C., 2003. From Local Challenges to a Global Community: Learning Reposi-tories and the Global Learning Repositories Summit. Version 1.0. Academic ADLCo-Lab, November 11.
EEE LOM, 2002. Draft Standard for Learning Object Metadata. IEEE Learning Tech-nology Standards Committee, IEEE 1484.12.1-2002, 15 July. (http://ltsc.ieee.org/wg12/files/LOM 1484 12 1 v1 Final Draft.pdf).
MS, 2007. Meta-data Best Practice Guide for IEEE 1484.12.1-2002 Standard for Learn-ing Object Metadata Revision: 31 August 2006, IMS Global Learning Consortium.Retrieved September 28, 2007, from http://www.imsproject.org/metadata.
raan, W., 2003. No One Standard Will Suit All. The Centre for Educational Tech-nology Interoperability Standards, Available http://www.cetis.ac.uk/content/20030513175232.
anouselis, N., Kastrantas, K., Tzikopoulos, A., 2007. An IEEE LOM applicationprofile to describe training resources for agricultural & rural SMEs. Proc.of the 2nd International Conference on Metadata and Semantics Research(MTSR”07), Corfu, Greece, October. (http://infolab-dev.aua.gr/files/publications/en/1200946986.pdf).
anouselis, N., Salokhe, G. (Eds.), 2008. Agricultural Learning Repositories (AgLR2008) E-Conference: Summary Report. Agricultural Learning Repositories TaskForce (AgLR-TF), June. (http://aglr.aua.gr/files/AgLR2008.pdf).
anouselis, N., Salokhe, G., Keizer, J., 2009. Comparing Different Metadata Applica-tion Profiles for Agricultural Learning Repositories. In: Sicilia, M.-A., Lytras, M.(Eds.), Metadata and Semantics, Metadata and Semantics, Springer, ISBN-978-0-387-77744-3, 469–479 (January).
cCormick, R., 2003. Keeping the pedagogy out of learning objects. In: SymposiumDesigning Virtual Learning Material, EARLI 10th Biennial Conference ImprovingLearning, Fostering the Will to Learn.
etros, S.E., Bennet, K., 2002. Learning Objects in Higher Education, Research Bul-letin. EDUCAUSE Center for Applied Research, November 1, 19.
iller, P., 1996. Metadata for the Masses, vol. 5. Ariadne.adolski, R., O’Neill, O., van der Vegt, W., Koper, R., 2006. Conformance testing the
Elixer within the chain for learning scenarios and objects. In: Proc. of the 6thIEEE International Conference on Advanced Learning Technologies (ICALT 2006),
Kerkrade, The Netherlands, 5–7 July, pp. 363–365.ajjar, J., 2008. Learning Object Metadata: An Empirical Investigation and LessonsLearned. PhD Thesis, Katholieke Universiteit Leuven, February.
ajjar, J., Ternier, S., Duval, E., 2003. The actual use of metadata in ARIADNE: anempirical analysis. In: Proceedings of the 3rd Annual Ariadne Conference, pp.1–6.
ics in Agriculture 70 (2010) 302–320
Najjar, J., Ternier, S., Duval, E., 2004. Interoperability of learning object repositories:complications and guidelines. IADIS International Journal of WWW/Internet.(http://www.cs.kuleuven.ac.be/∼najjar/papers/IadiswwwJournal.pdf).
Nelson, T., 1965. A file structure for the complex, the changing and the indeterminate.In: Proceedings of the ACM National Conference, pp. 84–100.
Neven, F., Duval, E., 2002. Reusable learning objects: a survey of LOM-based repos-itories. In: Proc. of the 10th ACM International Conference on Multimedia, pp.291–294.
NISO, 2004. Understanding Metadata. National Information Standards Organisation,NISO Press.
Ochoa, X., Duval, E., 2009. Quantitative analysis of learning object repositories. IEEETransactions on Learning Technologies.
Pisik, G.B., 1997. Is this course instructionally sound? A guide to evaluating onlinetraining courses. Educational Technology, 50–59, July–August.
PROLEARN, 2006. Research report 2005–2006. In: Wolpers, M. (Ed.), Network ofExcellence in Professional Learning. PROLEARN, June.
Retalis, S., 2004. Usable and interoperable e-learning resources and repositories. In:Sanjaya Mishra, Ramesh C. Sharma (Eds.), Interactive Multimedia in Educationand Training, Idea Group Inc., ISBN:1-59140-393-6.
Riddy, P., Fill, K., 2004. Evaluating e-Learning Resources. In: Proc. of the NetworkedLearning Conference, Lancaster, UK.
Roberts, A., 2003. Metaphysics of Metadata. Presentation, Available http://www.edusource.ca/english/resources/MetadataNB.ppt.
Sen, A., 2004. Metadata Management: past, present and future. Decision SupportSystems 37, 151–173.
Steinacker, A., Ghavam, A., Steinmetz, R., 2001. Metadata standards for web-basedresources. IEEE Multimedia (January–March), 70–76.
Stracke, C.M., Hildebrandt, B., 2007. Quality development and quality standards in e-Learning: adoption, implementation, and adaptation. In: Proceedings of WorldConference on Educational Multimedia, Hypermedia and Telecommunication2007 [= EdMedia 2007], AACE, Chesapeake, VA, pp. 4158–4165.
Stringer, R., 1992. Theseus: A project at Liverpool polytechnic to develop a hyper-media library for open and flexible learning. International Federation of LibraryAssistants, Attention Metadata Management 18 (3), 267–273.
Stuempel, H., Salokhe, G., Aubert, A., Keizer, J., Nadeau, A., Katz, S., Rudgard, S., 2007.Metadata application profile for agricultural learning resources. Proc. of the2nd International Conference on Metadata and Semantics Research (MTSR”07),Corfu, Greece, October. (ftp://ftp.fao.org/docrep/fao/010/ai154e/ai154e00.pdf).
Taylor, C., 2003. An Introduction to Metadata. University of Queensland Library,Available http://www.library.uq.edu.au/iad/ctmeta4.html.
Tzikopoulos, A., Manouselis, N., Costopoulou, C., Yialouris, C.P., Sideridis, A.B., 2005.Investigating digital learning repositories coverage of agriculture-related topics.In: Proc. of the International Congress on Information Technologies in Agricul-ture, Food and Environment (ITAFE05), Adana, Turkey, October.
Tzikopoulos, A., Yialouris, C.P., 2007. Knowledge repositories for rural communitiesof learning. In: Proc. of the 2nd European Conference on Technology-EnhancedLearning (EC-TEL 2007), Crete, Greece, September.
Tzikopoulos, A., Manouselis, N., Vuorikari, R., 2007. An overview of learning objectrepositories. In: Northrup, P. (Ed.), Learning Objects for Instruction: Design andEvaluation. Idea Group Inc., New York, pp. 29–55.
Wiley, D., 2000. Learning Object Design and Sequencing Theory. PhD Thesis, BrighamYoung University.
Wiley, D. (Ed.), 2002. The Instructional Use of Learning Objects. AECT, Bloomington,IN. Retrieved 10 September 2005 from http://reusability.org/read/.
International Conference on Computers and Advanced Technology in Education(CATE 2008), Heraklion, Crete, Greece, October.
Zschocke, T., Paisley, C., Duval, E., Beniest, J., 2005. CG Learning Object Metadata(LOM) Core, CGIAR ICT/KM OLR Project.
