通用不确定性推理模型

GENERALIZED UNCERTAINTY REASONING MODEL

不确定性推理是人类思维活动中最本质的东西,同时也是人工智能所研究的重要内容。已有的不确定性表示、度量以及推理方法,由于推理模型本身不完备,所考虑的不确定性因素是不全面的,只能适用于一些简单场合的推理。本文在全面剖析了产生式规则中各类命题的不同含义之后,提出了通用不确定性匹配算法、通用不确定性更新算法,形成了完整的通用不确定性推理(GUR)模型。GUR模型是非常完善的不确定性推理模型,首次区分开了六类命题的不同含义,澄清了以前的模糊概念。在通用不确定性匹配算法和通用不确定性更新算法中,引入了几个新函数,使推理过程更加清晰,更加合理。同时GUR模型也为进一步研究具体的不确定性推理方法奠定了基础。

Uncertainty reasoning is the essence of human thinking activities and is a key aspect of AI. Although the models of representation, measure and reasoning of uncertainty are being studied, the uncertainty factors that they touch upon are uncompleted because the reasoning model is immature. They only are suitable for some simple situations. In this paper, after analyzing the different meaning of various propositions, Generalized Uncertainty Match Algorithm (GUMA) and Generalized Uncertainty Renewing Algorithm (GURA) are advanced, and Generalized Uncertainty Reasoning (GUR) Model is formed. GUR model is a perfect uncertainty reasoning model. It distinguishes the different meanings of six propositions for the first time so that the former inklings are clarified. To make the reasoning process clearer and more rational, several new functions are introduced in GUMA and GURA. On the other hand, GUR model establishes a base of researching the concrete uncertainty reasoning methods.