Default Reasoning with Imperfect Information in Multivalued Logics

Abstract

The work provides an approach to default reasoning with imperfect information using extended logic programs in multivalued logics given by bilattices. The paper introduces the default semantics of an extended program using a set of operators with interesting properties elegantly expressed in terms of lattice theory, including the ultimate evaluation based consequence operator and the default operator corresponding to two types of inference in presence of imperfect information. We provide a computational approach for the default semantics and relate our framework to relevant work.