An approach for selecting appropriate methods for treating uncertainty in knowledge based systems

The problem of representation of and reasoning with uncertain data and knowledge is of importance in a broad range of disciplines, e.g. artificial intelligence and expert systems, decision theory and information systems development. The aim of this paper is to review the four developed uncertainty management systems (UMS), which are in most common use: Bayes (Probability) Theory, Fuzzy Logic, Certainty Factors Method and Dempster-Shafer Theory. The main features of each method are presented along with their strengths and weaknesses. A number of different sources of uncertainty are identified. The power of each of the four systems in dealing with these different types of uncertainties is examined. In the second part, a methodology for appropriate UMS selection is proposed. Selection is based on types of uncertainty inherent in a given application.