Uncertainty paradigms used in expert systems

This paper reviews four uncertainty paradigms used in expert systems, the Bayesian approach, certainty factors, Dempster-Shafer theory, and the theory of endorsements. The first three paradigms operate with quantitative measures of uncertainty; the last uses only qualitative expressions. Each paradigm is briefly discussed and strengths and weaknesses are noted. It is found that the mathematical rigor of the quantitative approaches may compromise their applicability in expert systems and that the qualitative approach lacks sufficient rigor. 6 refs., 1 tab.