Inference procedures under uncertainty for the problem-reduction method

From the viewpoint of efficient utilization of human knowledge in complex decision-making problems, the inference procedure under uncertainty is becoming more important for the problem-reduction method and expert systems. Unlike intuitive procedures employed so far in some expert systems, rational inference procedures are described in this paper on the basis of established Bayesian theory and Dempster and Shafer's theory of evidence. These results are extended to include fuzzy knowledge. As an alternative to the two probabilistic approaches which require idealized assumptions, fuzzy reasoning is introduced.