A machine learning algorithm for expert system based on MYCIN model

Abstract

What lead to the inaccurate certainty factor in MYCIN model were discussed in this paper, and a correction algorithm to adjust the certainty factor to the true one based on case studies was given, and knowledge representation model and inference method based on the certainty theory and the algorithm were analyzed, and the formula for certainty in reasoning process was given. Experiments proved that even certainty factors of rules of an expert system given at the beginning was inaccurate, by using the correction algorithm, the certainty factors could converge at actual values after adequate number of case studies.