Nonmonotonic reasoning and logic programming

摘要

Inference methods in knowledge-based systems are presented which are used for inferencing from incomplete information or in systems that model actions, i.e. systems that describe a time-varying world. In such cases, one is faced with nonmonotonic types of derivation when new information can lead to a revision of some statements accepted earlier. This paper presents a theoretical basis of an approach that incorporates Nute's notion of defeasible derivations. It allows updating of both knowledge items and derivations. Possible applications are discussed.