Answering queries on graphical decision models with fuzzy parameters: Algorithms and comparisons

Abstract

The reasoning strategy is a critical component for expert systems or decision support systems. After thorough self-evaluation and environmental scan, the knowledge workers need a reliable mechanism for diagnosis and an optimal suggestion for problem solving. This paper designs a reasoning model with a well-known graphical decision model, influence diagrams. In the proposed model, fuzzy probabilities as well as crisp probabilities are involved. The influence diagrams work compactly in problem diagnosis and decision making aid. Two algorithms to handle the fuzzy parameters, piecewise linearization and alpha-cut methods will be implemented and compared.