Issues of Integration and Balancing in Hybrid Knowledge Representation Systems

In the last several years the hybrid approach to Knowledge Representation has received much attention, because it was felt that one monolithic knowledge representation formalism cannot meet all representational demands. In this paper we will present one particular hybrid knowledge representation system, BACK, concentrating on matters of how to integrate different subformalisms and their interpretation. In particular, `balancing the expressiveness' of the respective subformalisms and combining the reasoning of the subsystems in a sound way is discussed. This will lead to a new view on the realization inference, first described by Mark, as a process of constraint propagation. 1 Introduction Aaron Sloman pointed out very clearly in [Sloman 85] the need for different knowledge representation formalisms for the adequate representation of a realistic portion of the world. This position contrasting the view that one uniform formalism is sufficient (e.g., [Kowalski 80]) is nowadays widely acc..