220 likes | 482 Views
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002. WordNet. „An On-line Lexical Database“ (Miller, G. A.; Beckwith, R.; Fellbaum, Chr.; Gross, D.; Miller, K. 1993, title). Based on psycho-linguistic insights (ibd.)
E N D
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 WordNet • „An On-line Lexical Database“ (Miller, G. A.; Beckwith, R.; Fellbaum, Chr.; Gross, D.; Miller, K. 1993, title). • Based on psycho-linguistic insights (ibd.) • „Nouns in WordNet: A Lexical Inheritance System“ (Miller, G. A. 1993, title). • „lexical resource“ (Guarino 1998, p. 12) • WordNet synsets are „[...] more like a thesaurus“ (Buitelaar 1998, p. 223) „Synset: A synonym set; a set of words that are interchangeable in some context.“ (WordNet Glossary of Terms. http://www.cogsci.princeton.edu/~wn/gloss) Is WordNet an ontology?
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 WordNet • „To be used as an ontology, however, some of WordNet’s lexical links need to be interpreted according to some formal semantics , (sic!) which tells us something about „the world“ and not (just) about the language. One of such links is the hyponym/hypernym relation, which corresponds in many cases to the usual subsumption (or IS-A) relation between concepts“ (Oltramari et al. 2002, p. 1 – own counting).
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 OntoClean • „First of all, we do not intend this [i.e. OntoClean – KMF] as a can-didate for a “universal” standard ontology. Rather, we support the vision of a library of foundational ontologies, reflecting diffe-rent commitments and purposes“ (Oltramari et al. 2002, p. 3 – ita-lics orig, own counting).
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 OntoClean • Ass.: „[...] [S]ubsumption [...] that is the basis of taxonomy is an extremely useful tool for imparting structure on an ontology. It is by far the most commonly used structuring primitive [...]“ (Guarino & Welty 2002, p. 63). • Impl.: „[...] these metaproperties impose constraints on the sub-sumption relation, which can be used to check the ontological consistency of taxonomic links“ (ibd., p. 62 – emph KMF). • „One of these constraints is that anti-rigid properties cannot subsume rigid properties“ (ibd.).
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 WordNet by OntoClean • Confusion between concepts and individuals i.e. sortals vs. instances or IS-A vs. INSTANCE-OF • E.g.: Territorial Dominion • The look of WordNet 1.6 original
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 WordNet
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 OntoClean
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 WordNet original • WordNet OntoClean WordNet original WordNet OntoClean
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 Apple in WordNet
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 WordNet
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 OntoClean
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 OntoClean • Ass.: „[...] [S]ubsumption [...] that is the basis of taxonomy is an extremely useful tool for imparting structure on an ontology. It is by far the most commonly used structuring primitive [...]“ (Guarino & Welty 2002, p. 63). • Impl.: „[...] these metaproperties impose constraints on the sub-sumption relation, which can be used to check the ontological consistency of taxonomic links“ (ibd., p. 62 – emph KMF). • „One of these constraints is that anti-rigid properties cannot subsume rigid properties“ (ibd.).
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 WordNet ! KMF
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 Apple in WordNetRedundancies/Ambiguities
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 WordNet Redundancy Ambiguity (resolved)
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 OntoClean
WordNet Psycho-linguistically motivated, implemented thesaurus May be used as an ontology subsumption-based Hyponyms Hypernyms only along with serious problems highly redundant huge inconsistencies Because of its ample and long-term generated database it is widespread. Besides: it is free. OntoClean Meta-Ontology Conglomerate of principles Examination tool for (existing) ontologies Purpose and Use Get a well-formed ontology Ontology and its entities/con-cepts/instances are: consistent coherent exclusively disjoint specific to a dominion exhaustive non-redundant, unambiguous and free of controversy Method Labelling of each entity by meta-properties. This causes a well-for-med ontology by use of sub-sumption axioms. University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 Conclusion
University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 References • Buitelaar, P.: CoreLex: An Ontology of Systematic Polysemous Classes. In: Guarino, N.: Formal Ontology in information systems. Proceedings of the First International Conference (FOIS '98), June 6-8, Trento, Italy. Amsterdam 1998: IOS. S. 221-235. • Guarino, N.: Formal Ontology and Information Systems. In: Guarino, N.: Formal Ontology in information systems. Proceedings of the First International Conference (FOIS '98), June 6-8, Trento, Italy. Amsterdam 1998: IOS. S. 3-15. • Guarino, N.; Welty, Chr.: Evaluating Ontological Decisions with OntoClean. In: Communications of the ACM #2, February 2002 (Vol. 45). Available on-line: http://www. ladseb.pd.cnr.it/infor/Ontology/Papers/OntologyPapers.html • Miller, G. A.: Nouns in WordNet: A Lexical Inheritance System. In: 5papers.pdf. Princeton 1993. Available on-line: ftp://ftp.cogsci.princeton.edu/pub/wordnet/5papers.pdf • Miller, G. A.; Beckwith, R.; Fellbaum, Chr.; Gross, D.; Miller, K. In: 5papers.pdf. Princeton 1993. Available on-line: ftp://ftp.cogsci.princeton.edu/pub/wordnet/5papers.pdf • Oltramari, A.; Gangemi, A.; Guarino, N.; Masolo; C.: Restructuring WordNet‘s Top-Level: The OntoClean approach. Padova 2002. Available on-line: http://www.ladseb.pd.cnr.it/infor/Ontology/Papers/OntologyPapers.html • WordNet Browser used: 1.6. Latest version is 1.7.1, available on-line for UNIX and Windows systems. Mac and DOS versions are not supported anymore. Free download of data set (also PROLOG version!), browser, documentation. http://www.cogsci.princeton.edu/~wn
OntoCleanAppendix University of Zurich • Computational Linguistics • Seminar „Wissensrepräsentation in der CL“ • Dr. Kai-Uwe Carstensen • SS 2002 • „First of all, we do not intend this [i.e. OntoClean – KMF] as a can-didate for a “universal” standard ontology. Rather, we support the vision of a library of foundational ontologies, reflecting diffe-rent commitments and purposes“ (Oltramari et al. 2002, p. 3 – ita-lics orig). • „Finally, we have to point out that the ontology presented here is an ontology of particulars. Properties and relations are therefore not part of its domain“ (ibd. – italics orig).