MMM-CADIAG: A mini modified CADIAG with mobius transform

Abstract

This paper describes the MMM-CADIAG system called Mini Modifid CADIAG with Moebius transform which is based on the algorithm of Mobius transform for CADIAG-2. This algorithm using Mobius trasform to compute new rule base for CADIAG-2. To apply Mobius transform for CADIAG-2 means to find new weights of fuzzy rules. This algorithm guarantees that using generalized MaxMin inference of CADIAG-2 the inference machine will reproduce the expert’s stated conditional beliefs as total degrees of confirmation and exclusion. MMM-CADIAG uses positive and negative knowledge. Knowledge Base of the system consists of set of IF - THEN rules. Each rule assigns its weight in [0, 1], With assumption that relative frequencies are used as weights of rules. MMM-CADIAG is able to calculate new weights of fuzzy rules and then suggest diagnoses by using the Mini Modified inference engine of CADIAG-2. Finally, a computer test program with established simple knowledge base in Oriental Medicine as an example is developed and tested. Programs is developed in C++ programming language and can run on PC/IBM computers.