ACM SIGART Bulletin - Special issue on implemented knowledge representation and reasoning systems
| Hi-index | 0.00 |
Terminological representation systems like BACK tended to focus on concept forming operators in their term description languages. In this paper I consider the integration of role forming operators for the definition of roles and its impact on role subsumption and hybrid entailment of facts. Examples show that defined roles considerably increase the expressiveness of the terminological component. Conseqences for concept subsumption are briefly discussed and then an algorithm computing subsumption between role terms is presented and shown to be turing reducible in polynomial time to subsumption between their domain and range concepts. More important, however, are the inferences resulting on the object level. I describe a recognition algorithm which derives the facts following from object descriptions and discuss issues of integration with existing recognizers. Since most of the entailed facts do not depend on subsumption of terms used in the object descriptions they are derived by the recognizer alone, without resort to the classifier. The presentation of the algorithms is intented to serve as a basis for implementations in hybrid representation systems.