摘要
本文讨论不精确推理合成算子中存在的问题.我们解决问题的方法是:研制一个系统,不应拘于一种理论,只采用这一理论中规定的合成算子,而应根据所合成问题的不同性质,采用符合三角模和三角余模的广义合成算子.我们提出一个补偿系数,根据它的大小,选择不同的合成算子,并把此选择问题列入知识获取过程中,以便确定的选择原则,符合专家的思维过程.
One of the essential problems in developing an expert sys(?)em is the uncertainty inference which we have to deal with meticulously.This paper discussed problems available in applying aggregation operators to uncertainty inference.By our point of view,it must not be limited by using those aggregative operators defined by that theory in a system.One can choose any kind of operators meeting the require- ments of T-norm and T-conorm according to properties of the matters which one need put into conjunction,disjunction etc.We suggested a compensation coefficent from which value we selected operators and the selected rule must be determined or discussed in the processses of knowledge acquisition in order to match the thought of human expert.
出处
《北方交通大学学报》
CSCD
北大核心
1989年第2期1-6,共6页
Journal of Northern Jiaotong University
关键词
专家系统
合成算子
不精确推理
knowledge engineering
expert system
uncertainty inference
