Affiliations: Dipartimento di Informatica e Sistemistica, Università di Roma “La Sapienza”, via Buonarroti 12, 00185 Roma, Italia
Note: [] This is an extended version of the paper “Utilizzo della Deduzione Terminologica in un Sistema Ibrido”, appeared in Proceedings of the First Conference of the Italian Association for Artificial Intelligence, Trento, Italy, 1989. Republication in english in AI Communications was suggested by the Programme Committee.
Abstract: The basic feature of Hybrid Knowledge Representation Systems is the separation of the knowledge represented within the system into distinct components. In particular, the architecture of several such systems includes two components: terminological, containing the knowledge about the classes of objects and their relationships, and assertional, with the knowledge about the individuals. The aim of this work is to study how the deduction methods developed for the terminological component can be effectively used in the hybrid deductions on the whole knowledge base. We consider a hybrid system, whose terminological component provides the user with a quite powerful language, while remaining computationally tractable, and whose assertional component simply allows one to express membership assertions between the individuals and the concepts of the terminology. In this paper, we present the languages for the two components, define the semantics of the whole knowledge base, and describe an efficient technique for the implementation of hybrid reasoning procedures, based on the notion of most specific concept.
