基于规则的专家系统诊断问题求解的一种模型

本文试图通过引进外推推理来克服基于规则的专家系统的不足,并为演绎推理和外推推理寻求一种一致的解释和理论基础.在讨论了规则的因果模型解释后,本文提出了诊断问题求解的最佳蕴含模型和基于该模型的循环生成-测试的推理方法.

A Theoretical Model for Diagnostic Problem-Solving in a Rule-Based Expert System

An attempt is made to overcome the defects of a rule-based expert system by employing the concept of abductive reasoning and to seek a uniform explanation and theoretical basis for abductive reasoning and deductive reasoning. The causal model explanation for rule is discussed. The Best Implication Model for diagnostic problem-solving and sequential hypothesis-and test reasoning are presented. A best implicative solution for diagnostic problem means that the solution is not only logically implicative, but also has a possibly large explanatory capacity, and is parsimonious. The reasoning based on the model is a sequential hypothesis-and-test process. Hypotheses constructed according to initial information lead the reasoning to a testing process in which new information is acquired. Hypotheses are confirmed, or abandoned, or revised depending on the new information. The sequential hypothesis-and-test process forms an ascending cycle, which is not ended until the explanatory capacity of the last hypothesis is satisfactory.