Automatic Ontology Extension: Resolving Inconsistencies

Abstract

Ontologies are widely used in text technology and artificial intelligence. The need to develop large ontologies for real-life applications provokes researchers to automatize ontology extension procedures. Automatic updates without the control of a human expert can generate potential conflicts between original and new knowledge resulting in inconsistencies occurring in the ontology. We propose an algorithm that models the process of the adaptation of an ontology to new information. 1 Automatic Ontology Extension There is an increasing interest in applying ontological knowledge in text technologies and artificial intelligence. Since the manual development of large ontologies proved to be a time-consuming task many current investigations are devoted to automatic ontology learning methods (see [6] for an overview). Several formalisms have been proposed to represent ontological knowledge. Probably the most important one of the existing markup languages for ontology design is the Web Ontology Language (OWL) based on the logical formalism called Description Logics (DL) [1]. In particular, description logics were designed for the representation of terminological knowledge and reasoning processes. Although most of the tools extracting or extending ontologies automatically output knowledge in the OWL-format, they usually use only a small subset of DL. The core ontologies generated in practice usually contain the subsumption relation defined on concepts (taxonomy) and general relations (such as part-of and others). At present complex ontologies making use of the whole expressive power and advances of the various versions of DLs can be achieved only manually or semi-automatically. However, several approaches appeared recently tending not only to learn taxonomic and general relations but also state which concepts in the knowledge base are equivalent or disjoint [5]. In the present paper, we concentrate on these approaches. We will consider only terminological knowledge (called TBox in DL) leaving the information about assertions in the knowledge base (called ABox in DL) for further investigations. 3 See the documentation at http://www.w3.org/TR/owl-features/