Journal: International Journal on Human-Computer Studies
Seiten: 945 - 965
Domain models can be constructed more easily and made more robust by reusing ontologies in a well-defined way. In this paper the KARO approach is introduced which provides various means of retrieving and adapting components of an ontology as part of a domain model construction process. KARO is based on the knowledge-processing component LILOG-KB provided by the LILOG text-understanding system. Above all, the notion of classification is applied for the retrieval of relevant categories. The upper structure of LILOG-KB serves as an exemplary ontology. By integrating KARO into the Model-based and Incremental Knowledge Engineering Environment (MIKE) the reuse of a predefined ontology can be integrated into the development process of expert systems in a systematic way.