Domain ontologies and wordnets in OWL: Modelling options

Abstract

Word nets are lexical reference systems that follow the design principles of the Princeton WordNet project (Fellbaum 1998, henceforth referred to as PWN1). Domain ontologies (or domain-specific ontologies, e.g. GOLD2 or the GENE Ontology3) represent knowledge about a specific domain in a format that supports automated reasoning about the objects in that domain and the relations between them (cf. Erdmann 2001, 78). Word nets have been used in various applications of text processing, e.g. discourse parsing, lexical and thematic chaining, cohesion analyses, automatic segmentation and linking, anaphora resolution, and information extraction. When these applications process documents dealing with a specific domain, one needs to combine knowlegde about the domain-specific vocabulary represented in domain ontologies with lexical repositories representing general vocabulary (like PWN). In this context, it is useful to represent and interrelate the entities and relations in both types of resources using a common representation language. In our research group “Text-technological Information Modelling4” we chose OWL as a common format for this purpose. Since our projects are mainly concerned with German documents, we developed an OWL model that relates the German wordnet GermaNet (henceforth referred to as GN)5 with domain-specific ontologies in an approach that was inspired by the Plug-In model proposed in Magnini/Speranza (2002). Our approach is decribed in Kunze et al. (to appear); it was evaluated using representative subsets of GN and of the domain ontology TermNet6 (henceforth referred to as TN) as data and Protégé