Hybrid inferences in KL-ONE-based knowledge representation systems

Abstract

We investigate algorithms for hybrid inferences in KL-ONE-based knowledge representation systems. Those systems employ two kinds of formalisms: the terminological and the assertional formalism. The terminological formalism consists of a concept description language to define concepts and relations between concepts for describing a terminology. On the other hand, the assertional formalism allows to introduce objects, which are instances of concepts and relations of a terminology. We present algorithms for hybrid inferences such as * determining subsumption between concepts * checking the consistency of such a knowledge base * computing the most specialized concepts an object is instance of * computing all objects that are instances of a certain concept.