Back to first pageBack to first page Centre for Artificial Intelligence of UNL
Browse our site

Semantic Diff as the Basis for Knowledge Base Versioning

Main informationBy: Ivan José Varzinczak (Meraka Institute)

Date: Wednesday, 28th of April 2010, 14h00

Location: FCT/UNL, Seminar Room (Ed. II)
AbstractIn this work we investigate the problem of maintaining and reasoning with different versions of a knowledge base. We are interested in the scenario where a knowledge base (expressed in some logical formalism) might evolve over time and, as a consequence, different versions thereof have to be maintained simultaneously in a parsimonious way. Moreover, users of the knowledge base should be able to access, not only any specific version, but also the differences between two given versions of the knowledge base. We address this problem by proposing a general semantic framework for the maintenance of different versions of a knowledge base. It turns out that the notion of semantic difference between knowledge bases plays a central role in the framework. We show that an appropriate characterization produces a unique definition of semantic difference which is applicable to a large class of logic-based knowledge representation languages. We then proceed to restrict our attention to finitely generated propositional logics, and show that our semantic framework can be represented syntactically in a particular kind of normal form, referred to as ordered complete conjunctive normal form or oc-CNF. This is followed by a generalization in which we show that similar results can be obtained for any syntactic representation (in a finitely generated propositional logic) of the semantic framework. Of particular interest are representations of appropriately chosen normal forms. We expect that our constructions for the propositional case can be extended to more expressive languages, such as description logics (DLs). In that respect, our results add to the investigation of the versioning problem for DL-based ontologies.
(Joint work with Enrico Franconi and Thomas Meyer.)
Short-bioIvan José Varzinczak received his PhD in Artificial Intelligence from Université Paul Sabatier, France, in 2006. Former member of the LILaC research group at IRIT, he is currently a postdoc researcher at Meraka Institute in Pretoria, South Africa. He received his MSc (2002) and BSc (2000) in Computer Science from the Federal University of Paraná, Brazil.
Ivan is author of articles in journals (JAIR, AI Journal, JANCL) and papers in conferences (IJCAI,KR, ECAI, JELIA, AiML) and in refereed workshops (NMR, Commonsense, NRAC, M4M). He is PC member of ECAI’10, and co-chair of the IJCAI’09 and ECAI’10 workshop series ARCOE, and also co-chair of the Commonsense and NMR for Ontologies sub-workshop of NMR’10.
Ivan’s main research interest areas are logic-based knowledge representation and reasoning, with focus on non-monotonic reasoning, modal logics and description logics. He has been working on belief change and reasoning about actions. Among his main current research topics are pertinence logics, Horn belief change, action theory change, and ontology repair.

Centre for Artificial Intelligence of UNL
Departamento de Informática, FCT/UNL
Quinta da Torre 2829-516 CAPARICA - Portugal
Tel. (+351) 21 294 8536 FAX (+351) 21 294 8541

Fundacao_FCT