Terminologies and rules

Abstract

Terminological logics have become a well understood formal basis for taxonomic knowledge representation, both for the semantics (classically by Tarski models) and for the inference services (like concept subsumption, instantiation, classification, and realization) of terminological systems of the KL-ONE family. It has been demonstrated that terminological reasoning can be realized by efficient and logically complete algorithms based on tableaux style calculi. However, representation of and reasoning with terminological information supports just a rather static form of knowledge representation. Only a fixed description of a domain can be represented: There is the schematic description of concepts in the socalled TBox and the instantiation of concepts by individuals and objects in the ABox of such systems. Terminological inferences can retrieve implicit information, but cannot be used for deriving new data. In order to overcome this restriction terminological systems often allow for additional rule based formalisms. Those, however, are missing a clear declarative semantics. In this paper we will sketch several declarative forms of rule based extensions of terminological systems that have been developed recently.