A fuzzy distributed expert system design tool EXGEN and its application tosensory foodstuff evaluation

Since the appearance of MYCIN, expert systems have been widely and successf ully developed for various scientific and technological researches and appl ications. These applications require more and more fuzzy information resour ces because of the uncertainty, inexactness in labeling facts using linguis tic terms and expressing human expertise. Sensory foodstuff evaluation is a mong this kind of fuzzy expert system applications. In the frame of the research project on fuzzy expert systems for science an d technology at the Hanoi University of Technology, we have developed an ex pert system building tool called EXGEN which has the following features: Knowledge editing in the form of production rules using Vietnamese in the n atural language-like syntax. The tool is also capable to verify the consist ency of an acquired knowledge base. Inference engine consisting of two principal inference mechanisms (forward and backward inference) and control strategy module. We proposed also some heuristics for choosing a potential inference trace, allowing to get more i nformation about conclusions. Possibility of establishing a configuration for a distributed working sessi on. It would be possible to carry out: + a deduction over a shared rule base (RB) in the server, based on informat ion acquired from workstations (common RE and conclusion, distributed fact base (FB)) + a deduction over a shared RE in the server with different cognitive tasks (including hypotheses fact and conclusions) on workstations (common RE and distributed FB) + deductions on workstations with distributed knowledge bases (Distributed RE and FB) We have already implemented an application expert system SENEXSYS for senso ry foodstuff evaluation using the building tool EXGEN. Experimental results have shown that qualification given by the expert system is comparable to evaluation results obtained by following up Vietnamese standard TCVN 3215.7 9.