智能DSS中非精确推理可信度的测宽研究

在智能决策支持系统（ＤＳＳ）研究中，知识表示、自然语言理解以及相应的推理系统是急待解决的几大问题。传统的精确推理已难以适应计算机应用发展的需要。近年来在推理机的设计中发展了规则与结论都不唯一的非精确、非单调推理。但如何测定结论的可信度、可用度仍是个问题。本文探讨基于信息论引入测度的概念，以解决非精确推理结论的可信度问题。

The Measure Research of Likelihood Reasoning Credibility in the IDSS

In the research of the intelligence decision support systems the major problems vexingexperts are knowledge expressing, natural language understanding and assodat.ed reasoning system.The conventional pacely reasoning system can hardly match the development of computerscience. However, some theories have been proposed to work out the dilemma recently.They are likelihood Reasoning and Nonmontonic Reasoning. But how to measure the credibility,practicibility and how to quantify the likelihood remain to be queshons.We know that it is impossible to get the relevance of complicated case out of simple case.We try to circumspectthe problem on the basis of Information Theory. And in this paper we propose some rules to improve the reasoning system.