Research and Application of Uncertain Inference in Expert Systems

Expert system is an important part of Artificial Intelligence, which has realized a significant breakthrough of the Artificial Intelligence: From its theory research to application, and from its general thought method discussion changed over to use the specialized knowledge to solve the special problems. All kinds of expert systems have been widely used in industry, agriculture, finance, military, and so on, which have made the huge economic efficiency and social efficiency.In the actual process of researching and developing the expert system, because of the solution to the expert system question does not need to be serious and accurate like classical discipline such as mathematics and physics. The domain expert’s knowledge and the information we need to deal with are always uncertain, imprecise, incompletely knows, but are fuzzy, incomplete, inconsistent and changes. Therefore, in order to express these indefinite knowledge in the expert system, and can use these indefinite knowledge which formalized to carry on the judgment, the inference and the decision-making, that we must not only study the indefinite and the imprecise knowledge expression method, but also must discuss the imprecise inference method. There are lots of indefinite facts in this world, and the relations between facts and conclusions are not always inevitable, they are a fuzzy set. The human experts often could infer the almost reasonable conclusion by using the indefinite fact and the causal relation. Only if introduce the mechanism of imprecise inference into computer, and make it simulates human’s inference decision-making process, the expert system can become the reality feasible system. It can be said that the use of imprecise inference in the expert system design is inevitable, and it becomes an important question which related to the success or not of expert system design. So, imprecise inference model is the core of the expert system.