Upload
lehanh
View
216
Download
0
Embed Size (px)
Citation preview
1
Ontologies
Ana Paula Rocha
Electronic Business Technologies
TNE
Motivation
Battery
– Different features
– Different prices
– Different utilities
2
2
TNE
Ontology
• Origin in philosophy: – Specification of what exists or what we can say about the world
• In AI systems:– what "exists" is what can be represented
• Popular topic since the early ninety
• Several communities of AI research:– Knowledge Engineering/Representation– Natural Language Processing– Intelligent Information– Information Retrieval on the Web
3
TNE
Ontology
• Reason for the popularity: mainly due to the promise of a shared and common understanding of some domain of knowledge that can be communicated between people and computer
4
3
TNE
What is an ontology?
• “An ontology defines the basic terms and relations comprising the vocabulary of a topic area as well as the rules for combining terms and relations to define extensions to the vocabulary” [Neches et al., 1991]
• “An ontology may take a variety of forms, but necessarily it will include a vocabulary of terms, and some specification of their meaning. This includes definitions and an indication of how concepts are inter-related which collectively impose a structure on the domain and constrain the possible interpretations of terms” [Uschold e Jasper, 1999]
5
TNE
What is an ontology?
• “is an explicit specification of a conceptualization” [Gruber, 1993]
– Conceptualisation: is a set of definitions that allows one to construct expressions about some physical domain.
– Explicit: means that the concepts and relationships of the abstract model are given explicit terms and definitions.
• “is a formal specification of a shared conceptualization”[Borst, 1997]
– Formal: Refers to the fact that an ontology should be machine-readable.
– Shared: reflects the notion that the ontology captures consensual knowledge, that is, it is not the privilege of some individual, but accepted by a group.
6
4
TNE
What is an ontology?
• The ontology community distinguishes between: ontologies that are mainly a taxonomy; and ontologies that model the domain more deeply providing more constraints on the semantics of the domain
Lightweight:– make scarce or no use of axiomsto model knowledge and clarify the meaning of concepts in the domain. – include concepts, relationships between concepts and properties that describe these concepts.
Heavyweight:– make intensive use of axiomsto model knowledge and restrict domain semantics. – add axioms and restrictions to lightweightontologies
7
TNE
Ontology
Issues to discuss:
• Ontology construction– methodology
– tools
– languages
• Ontology Learning
• Ontology Mapping
• Ontology Translation and Interoperability
• Applications
8
5
TNE
Ontology construction
Language?
Tools?Methodology?
9
TNE
Ontology construction
• Developing an ontology involves (basically):– Define domain and scope
Example– Domain: Wine representation
– Scope: applications that suggest combinations of wines and food
Other scopes:– Helping clients in the restaurant to decide which wine to ask
– Helping buyers of wine
– Helping transactions between wine producer and wine reseller
10
6
TNE
Ontology construction
• Define classes in the ontology
• Organize classes in a taxonomy (subclass-superclass)
11
WineWhite WineRed WineRose Wine
Red Bordeaux
Red Burgundy
Producer
TNE
Ontology construction
12
• Define attributes (slots)
color
alcoholic content
taste
shape
7
TNE
Ontology construction
• Define relations
13
Producer
produces {wine}
TNE
Ontology construction
• Define instances: elements
• Define axioms: sentences that are always true
• Define functions: example, price calculation
• ….
14
8
TNE
Ontology construction
Questions about methodology, tools and languages:
– What methodologies are available for building ontologies, or to reuse existing ontologies?
– What is the life cycle of an ontology?
– What tools support the process of developing an ontology?
– What language should be used?
– Which expressivity has a language of ontology?
– The language chosen is appropriate for the exchange of information between different applications?
15
TNE
Methodologies for Ontology construction
• Enterprise Ontology
• TOVE (Toronto Virtual Enterprise)
• METHONTOLOGY
• On-To-Knowledge
16
9
TNE
Methodologies for Ontology construction
Enterprise Ontology- Uscholdand King’s Method[Uschold e King, 1995]
17
Identify the ontology proposal
Build the ontology: capture, codify and integrate appropriate knowledge from existing ontologies
Evaluate the ontology
Document the ontology
TNE
Methodologies for Ontology construction
TOVE (Toronto Virtual Enterprise) [Grüninger e Fox,1995]
18
Identify motivationscenarios
Formulate questions to answer
Formal terminology
Formulate questionsin FOL
Specify axioms
Evaluate theontology
10
TNE
Methodologies for Ontology construction
Methontology[Gómez-Pérez, 1998]
– Specify requirements
– Conceptualize the domain of knowledge
– Formalize the conceptual model in a formal language
– Implement a formal model
– Maintain the implemented ontologies
• Activities performed during the construction process : – Knowledge acquisition
– Integration
– Evaluation
– Documentation
19
TNE
Methodologies for Ontology construction
On-To-Knowledge[Staab et al., 2001]
20
Refinement
Evaluation
Maintenance
11
TNE
Methodologies for Ontology construction
Conclusions:
• Methontology: – recommended by FIPA (Foundation for Intelligent Physical Agents)
• Proposals not unified: – each group applies its own approach
21
TNE
Tools for Ontology construction
22
Ontolingua WebONTO WebODE
Protégé OntoEdit OilEd
Apollo SymOntoX OntoSaurus
DagEdit DOE IsaViz
SemTalk OntoBuilder DUET
12
TNE
Tools for Ontology construction
Protégé[Noy et al., 2000]
http://protege.stanford.edu
– Developed by the Medical Informatics group, Stanford University
Main features:– Open Code
– Standalone application
– Extensible architecture
– Ontology Editor + plugins (library of functionalities)
– Currently imports/exports to Flogic, Jess, OIL, XML, Prolog, OKBC access
23
TNE
Tools for Ontology construction
24
13
TNE
Tools for Ontology construction
WebODE[Arpírez et al. 2001; Corcho et. al, 2002]
http://delicias.dia.fi.upm.es/webODE
– Developed by the Artificial Intelligence Laboratory, Technical University of Madrid
Main features:– Extensible architecture
– Web application
– Import/export to XML, RDF(S), OIL, DAML+OIL, CARIN, Flogic, Jess, Prolog
– Ontologies stored in relational databases
– Documentation services, evaluation services and merging of ontologies
25
TNE
Tools for Ontology construction
• OntoEdit[Sure et al., 2002]
http://ontoserver.aifb.uni-karlshure.de/ontoedit
– Developed by AIFB (Institutf ür Angewandte Informatik und Formale Beschreibungsverfahren), University of Karlsruhe
Main features:– Extensible architecture, based in plugins
– Import/export for Flogic, XML, RDF(S), DAML+OIL
– Two versions: OntoEditFree e OntoEditProfessional
26
14
TNE
Tools for Ontology construction: comparison
• Expressiveness:– All the tools allow to represent classes, relations, attributes, instances and
axioms.
• Interoperability– Many of the tools import and export for XML and markup languages.
– There is no study on the quality of translators.
– There is no results on the exchange of ontologies between different tools.
• Methodology– WebODE supports Methontology
– OntoEdit supports On-To-Knowledge
• Cooperative and Collaborative Ontology Construction – WebODE has the more advanced features
27
TNE
Languages for Ontology construction
Traditional ontology languages
Cycl Ontolingua F-Logic CML OCML Loom KIF
Standard languages for Web
XML RDF
Web-based ontology languages
OIL DAML+OIL SHOE XOL OWL
28
15
TNE
Ontology Learning
• Ontology learning, set of methods and techniques used for:– building an ontology from scratch
– enriching, or adapting an existing ontology in a semi-automatic fashion using several sources
• Several approaches exist, using sources like: – texts, instances, databases schemes, XML schemes, ...
• The most widely used and interesting in the Semantic Web context is the approach based on texts
• Ontology Learning from texts:– extract ontologies by applying natural language analysis and machine
learning/linguistic techniques
29
TNE
Ontology Learning: tools
Tools based on natural language analysis and machine learning techniques:
– Conceptual Clustering, concepts are grouped according to the semantic distance between each other to make up hierarchies.
ASIUM [Faure e Nedellec, 1999], Mo´K [Bisson et al., 2000] e SVETLAN [Chaelandar e Grau, 2000]
– Lexical and Syntatic Analysis
Corporum-Ontobuilder [htt://ontoserver.cognit.no/], LTG [Mikheev e Finch, 1997] e Terminate [Biébow e Szulman, 1999]
– Statistical Approach
Text-To-Onto [Maedche e Staab, 2001]
30
16
TNE
Ontology Mapping
• Ontology Mapping can be defined as a function that associates terms and expressions defined in a source ontology with terms and expressions defined in a target ontology
Main tools: – Chimaera, PROMPT, OBSERVER, OntoMorph, Auto-Categorizer,
WebPicker
31
TNE
Ontology Mapping
• Is a difficult task because:– it requires a thorough verification of inheritance, consistency of
inference, ...
– the relationships can be many-to-one, one-to-many, many-to-many, within a domain or across domains.
• Many tools are limited to:– verify classes or relations
– check consistency
– provide a list of recommendations of what to do
32
17
TNE
Translation and Interoperability
• Ontologies are built using different languages– Each language has its syntax, expressiveness and reasoning ability
– based on different paradigms (frames, first-order logic, description logic, etc.)
• Ontologies are built using different development tools– Each tool exports/imports ontologies for one or more languages
33
TNE
Translation and Interoperability
• Translation problem – arises when we decided to reuse an ontology or part of it, with a tool or
language different from that in which the ontology is available.
• Ontology tools should be able to:– exchange ontologies between them
– export/import ontologies in different formats
• If we refer to the exchange of ontologies between different tools, the problem of translation is also known as interoperability between ontology tools
34
18
TNE
Translation and Interoperability
• An initial proposal was:– use KIF as a format of knowledge exchange in the Ontolingua Server
– which would reduce the number of translators to be developed
Ontolinguaprovides a distributed collaborative environment to browse, create, edit, modify, and use ontologies(http://ksl.stanford.edu/software/ontolingua)
• Proposal failed: – very poor translation quality
– facilities for export but not import, each developer had to build their own translators to Ontolingua and KIF
• New tools for ontology construction– have created their own translators for different languages
35
TNE
Translation and Interoperability
Conclusions
– Problem of translation has not been addressed in an integrated way. Integrated means:
• examine in depth all the problems that appear in translations
• propose theoretical solutions to these problems
• simultaneously provide technological solutions to solve the problems
– No current proposal addresses the problem of the loss of information in translation
36
19
TNE
Ontology: applications
• Knowledge Management– integration of heterogeneous, distributed and semi-structured
information resources
• Electronic Commerce– business relationships (buy/sell) between business entities (especially
B2B)
– places such as Yahoo organize your content into categories to help users to navigate
– the United Nations Standard of Products and Services Code<http://www.unspsc.ORG/> contains a taxonomy that organizes products and services to facilitate transactions between B2B sites that agree with the vocabulary defined there (ontological commitment)
37
TNE
Aplicações de Ontologias
• Intelligent Information – Search engines like Google and AltaVista use ontologies to implement
semantic queries that improve the classic search by keyword.
• Natural Language Processing– Ontologies like WordNet are used to represent grammatical structures
that allow to perform semantic analysis of texts by reducing the ambiguity of natural language semantics. http://www.cogsci.princeton.edu/~wn/
• Enterprise Modelling – Ontologies support the organizational memory of an enterprise, that
allows the interoperation of departments/areas by using a common vocabulary and pre-defined rules. Examples of these ontologies can be found in TOVE and The Enterprise Ontology.
38
20
TNE
Conclusions
There is a need to work on creating tools that facilitate:
– Ontology development throughout all the life cycle, including: integration, merging, reengineering, content evaluation, translation into different languages and formats, and content exchange with other tools
– Ontology management: configuration and ontology evolution management
– Ontology support: schedule, documentation, advanced techniques for viewing the contents of the ontology, etc.
– Methodological support for building ontologies
39
TNE
Some bibliography• [Arpírez et al., 2001] Arpírez JC, Corcho O, Fernández-López M, Gómez-Pérez A. WebODE:
a scalable ontological engineering workbench. In: Gil Y, MusenM, ShavlikJ (eds) First International Conference on Knowledge Capture (KCAP’01). Victoria, Canada. ACM Press (1-58113-380-4), New York, pp 6-13.
• [Biébow e Szulman, 1999] Biébow B, Szulman S. TERMINAE: a linguistic-based tool for the building of a domain ontology. In EKAW’99 –Proceedings of the 11th European Workshop on Knowledge Acquisition, Modellingand management. Dagstuhl, Germany, LCNS, pages 49-66, Berlin, 1999. Springer-Verlag.
• [Bisson et al., 2000] Bisson G, Nedellec C, Cañamero D. Designing Clustering Methods for Ontology Building –The Mo’KWorkbench. In S. Staab, A. Maedche, C. Nedellec, P. WiemerHasting(eds.), Proceedings of the Workshop on Ontology Learning, 14th European Conference on Artificial Intelligence, ECAI’00, Berlin, Germany, August 20-25.
• [Borst, 1997] Borst WN. Construction of Engineering Ontologies. University of Tweenty. Enschede, The Netherlands -Centre for Telematicaand Information Technology.
• [Chaelandar e Grau, 2000] Chaelandar G, Grau B. SVETLAN’-A System to ClassigyWords in Context. In S. Staab, A. Maedche, C. Nedellec, P. Wiemer-Hastings (eds.) Proceedings of the Workshop on Ontology Learning, 14th European Conference on Artificial Intelligence ECAI’00, Berlin, Germany, August 20-25.
40
21
TNE
Some bibliography
• [Chalupsky, 2000] Chalupsky H. OntoMorph: a translation system for symbolic knowledge. In: Cohn AG, GiunchigliaF, Selman B (eds) 7th International Conference on Knowledge Representation and Reasoning (KR’00). Breckenridge, Colorado. Morgan Kaufmann Publishers, San Francisco, California, pp 471–482.
• [Corcho e Gómez-Pérez, 2001a] Corcho O, Gómez-Pérez A. WebPicker: Knowledge Extraction from Web Resources. 6th Intl. Workshop on Applications of Natural Language for Information Systems (NLDB'01). Madrid. June, 2001.
• [Faure e Nédellec, 1999] Faure D, Nédellec C. Knowledge acquisition of predicate argument structures from technical texts using machine learning: The system ASIUM. In D. Fenseland R. Studereditors, Proc. Of the 11th European Workshop (EKAW’99), LNAI 1621, pages 329-334. Springer-Verlag.
• [Gómez-Pérez, 1998] Gómez-Pérez A. Knowledge Sharing and Reuse. In: LiebowitzJ (ed) Handbook of Expert Systems. CRC Chapter 10.
• [Gruber, 1993] Gruber TR. A translation approach to portable ontology specification. Knowledge Acquisition 5(2)199–220.
• [Grüninger e Fox, 1995] Grüninger M, Fox MS. Methodology for the design and evaluation of ontologies. In: IJCAI95 Workshop on Basic Ontological Issues in Knowledge Sharing. Montreal, Canada.
41
TNE
Some bibliography• [McGuinness et al., 2000b] McGuinness DL, Fikes R, Rice J, Wilder S. An environment for
merging and testing large ontologies. In. Proc. 7th Intl. Conf. On Principles of Knowledge Representation and Reasoning (KR2000), Colorado, USA, April 2000.
• [Maedche e Staab, 2001] Maedche A, Staab S. Ontology Learning for the Semantic Web. IEEE Intelligent Systems, Special Issue on the Semantic Web, 16(2).
• [Mikheev e Finch, 1997] Mikheev, A. Finch, S. A Workbench for Finding Structure in Texts. Proceedings of ANLP-97 (Washington D.C.). ACL March 1997. pp 8.
• [Neches et al., 1991] Neches R, Fikes RE, Finin T, Gruber TR, Senator T, Swartout WR. Enabling technology for knowledge sharing. AI Magazine 12(3):36–56.
• [Noy e Musen, 2000] Noy NF, Musen MA. PROMPT: Algorithm and Tool for Automated Ontology Merging and Alignment. In: 17th National Conference on Artificial Intelligence (AAAI’00). Austin, Texas.
• [Staab et al., 2001] Staab S, Studer R, Schnurr HP, Sure Y. Knowledge Processes and Ontologies. IEEE Intelligent Systems, 16(1) (2001).
• [Sure et al., 2002] Sure Y, Erdmann M, Angele J, Staab S, Studer R, Wenke D. OntoEdit: Collaborative Ontology Engineering for the Semantic Web. In: HorrocksI, HendlerJ (eds) First International Semantic Web Conference (ISWC’02). Sardinia, Italy. Springer VerlagLecture Notes in Computer Science (LNCS) 2342. Berlin, Germany, pp 221–235.
42
22
TNE
Some bibliography
• [Uschold e Jasper, 1999] Uschold M, Jasper R. A Framework for Understanding and Classifying Ontology Applications. In: BenjaminsVR (ed) IJCAI'99 Workshop on Ontology and Problem Solving Methods: Lessons Learned and Future Trends. Stockholm, Sweden. CEUR Workshop Proceedings 18:11.1–11.12. Amsterdam, The Netherlands (). http://CEUR-WS.org/Vol-18/
• [Uschold e King, 1995] Uschold M, King M. Towards a Methodology for Building Ontologies. In: IJCAI’95 Workshop on Basic Ontological Issues in Knowledge Sharing. Montreal, Canada.
43
Ontology Services
23
TNE
Electronic Institution
45
Norms & Rules
Electronic Institution
links to other institutions
legal
financial
VE
Formation
Q-
Negotiation
VE
Operation
Monitoring
VE
Dissolution
Ontology Services
Electronic ContractMAgt EAgtEAgt EAgt
Trust & Reputation
TNE 46
Ontology
• Common and shared understanding about a domain
• Agents can use ontologies that are not exactly equal to represent their vision of the domain
• Institutional Ontology– defines a business vocabulary to be used by all agents
– includes: Concepts, AgentActions, Predicates
24
TNE 47
Institutional Ontology
TNE 48
Interoperability problem
• In a decentralized and distributed environment, interoperability refers to how the communication takes place between humans and software agents.
• Ontologies are developed by different and heterogeneous people and continue to evolve over time
25
TNE 49
Ontology services
In solving the interoperability problem in e-commerce, particularly in B2B transactions, some ontology services are particularly useful:
– Definition of attributes’ dependencies for each product
– Translation of terms between two ontologies for the same domain
– Conversion of values (eg different metrics)
– Report on mandatory or different attributes that are under negotiation
Ontology-based Services Agent
present in the Electronic Institution
TNE 50
Ontology Services Agent (OSAg)
• The Electronic Institution integrates(among others) anOntology–based Services Agent(OSAg)
• OSAg offers the following services:– Matching terms– Conversion of units
• Matching terms– when an agent does not understand the contents of a message– based on lexical and semantic similarity measures
• comparison of attributes, relationships between concepts, and concepts descriptions
26
TNE 51
Matching terms (OSAg)
• Syntactic Similarity between attributes– Calculates a comparison value ”3-gram”
• Syntactic Similarity between descriptions– Only used the most representative words– A "3-grams” matrix is calculated between each word of description– is used the formula rn-grams
nr
n
ii
gramsn
∑=
− = 1max
n
rsim gramsn
attrSetattrSet∑ −=2/1
for each data type: string, integer, float, boolean, has-part
Maxi is the maximum of all comparison results that exist for one attribute type
for all data types
TNE 52
Matching terms (OSAg)
• Semantic Similarity– Semantic Similarity mesure LCH (Leacock-Chodorow), based on
“WordNet”
• Final Similarity value
– weak correspondence (0.55 – 0.59)
– approximate correspondence (0.6-0.69)
– strong correspondence (0.7 – 1.0)
3
3
12/1
∑=
×= i
imethod
termterm
weightingresultsim
i
27
TNE 53
Ontologia
• Ontology Services
– Institutional ontology defines a business vocabulary
– Ontology-based Services Agent solves the interoperability problem
TNE 54
Relé
Fusível
Vedante Rolamento
Parafuso
comprimento : DoublecabecaParafuso : String = hexagonal, allel, estrelada
Porcas
largura : DoubletipoFreio : String = plático, nylon, mecânico, sem freio Anel
largura : Double
Caixa Direção
percurso : Double Bomba Direção
FarolFrente FarolAtras
Sistema
tipo : String = monotronic, k-jetronic, mono-jetronic
Especif icacaoMotor
descricao
RodasDentadas
numeroDente : IntegertipoDente : String = reto, espiral, cônico
Pneu
largura : DoublerelacaoAspecto : IntegerdiametroInterior : Double
Disco
diametroInterno : Doublelargura : DoublenumFuros : Integermaterial : String = liga de aço, alumínio, aço
Transmissões
comprimento : Doublediametro : DoubletipoSistemaRotula : String = por bolas, por cruzes, de agulhas
ABS
AirBag
SegurançaPassiva
Bomba
pressaoTrabalho : DoubletipoBomba : String = pistões, embulo, ...
DiscoTravão
arqProjeto : arquivo
TuboTravagem
arqProjeto : arquivo
Corrente
largura : Doublecomprimento : DoubletipoDentes : String = quadrado, redondo, trapéziotipoBorracha : String
Cabo da Corrente
comprimento : Doubleresistencia : Integermaterial : String = LISTA
Eletrônica
11
11
Relé-Fusivel
intensidade : IntegersistemaFuncionamento : String = contador, ruptor
Vela
resistencia : Integerdiametro : Doubleassento : String = cônico, anel
Engrenagem
1..n1..n
Transmissão
1..n1..n
1. .n1. .n
1..n1..n
Outros
quantidade : IntegerdiâmetroNominal : String
Farol
potenciaEletrica : IntegernumeroLampada : Integercor : String
Janela
altura : Doublecomprimento : Doubleespessura : DoubletipoCristal : String = laminado simples, laminado duploformato : StringarqProjeto : arquivo
BorrachaVedante
comprimento : DoubletipoBorracha : String = macia, oca, ...
TuboBorracha
diametroInterior : Doublecomprimento : DoublepressaoMaxima : Double
Tinta
cor : Integerkg : Double
SistemaSegurança
arqProjeto : arquivo
1..n1..n
1..n1..n
1..n1..n
SistemaTravagem
1..n1..n
1. .n1. .n
1..n1..n
Vedante-Rolamento
diametroExterior : DoublediametroInterior : DoubleExpessura : DoublenumRotacao : Integer
Motor
11
1..n1..n
1..n1..n
1..n1..n
1..n1..n
Caixa
11
Automovel
conceito
1..n1..n
1..n1..n
1..n1..n
11
1..n1..n
1..n1..n
11
11
11
Componente
quantidade : IntegerpressaoTrabalho : Doublealimentacao : String = mecânica, elétrica
Direção
11
11
Sinônimo
descriçãoSinônimo : String
Sinonimizavel
1..n1..n 1..n1..n
temTodas as classes, com excessão das subclasses herdam da classe Sinonimizavel. Nem todos os relacionamentos foram colocados para não dif icultar a visualizaçãoe leitura do diagrama.
UMLspecification for the “car” ontology (example)