A Hybrid Framework for Representing Uncertain Knowledge

Abstract

This paper addresses the problem of bridging the gap between the fields of Knowledge Representation (KR) and Uncertain Reasoning (UR). The proposed solution consists of a framework for representing uncertain knowledge in which two components, one dealing with (categorical) knowledge and one dealing with uncertainty about this knowledge, are singled out. In this sense, the framework is "hybrid". This framework is characterized in both model-theoretic and prooftheoretic terms. State of belief is represented by "belief sets", defined in terms of the "functional approach to knowledge representation" suggested by Levesque. Examples are given, using first order logic and (a minimal subset of) M-Krypton for the KR side, and a yes/no trivial case and Dempster-Shafer theory for the UR side. 1. Introduction An impressive work has been carried out over the last two decades in the fields of Knowledge Representation (KR) and of Uncertain Reasoning (UR), resulting in a number of concepts being inv...