Controlling expert system recommendations with defeasible logic

Abstract

Nonmonotonic logics are alternatives to probabilistic systems for reasoning with uncertain or incomplete information. For the most part, these logics have not been implemented as automated reasoning systems or incorporated into expert systems that deal with real problems. Defeasible logic is a nonmonotonic formalism implemented in an extension of the Prolog logic programming language called d-Prolog. FORE is an initial prototype expert system for selecting a business forecasting method. Written in d-Prolog, FORE demonstrates the feasibility of using defeasible logic to control the recommendations of expert systems with a significant degree of complexity. It also serves the secondary purpose of providing a foundation for development of a mature forecasting method selection system.