条件事件代数研究综述  

A Review on Theory of Conditional Event Algebra

作　　者：邓勇[1] 刘琪[2] 施文康[1] 

机构地区：[1]上海交通大学电子信息学院,上海200030 [2]上海交通大学生命科学技术学院,上海200030

出　　处：《计算机学报》2003年第6期650-661,共12页Chinese Journal of Computers

摘　　要：综述了条件事件代数理论的原理、主要性质和应用 .条件事件代数是一门新兴的解决不确定性、概率性和模糊性推理问题的学科 ,是在确保规则概率与条件概率相容的前提下 ,把布尔代数上的逻辑运算推广到条件事件(规则 )集合中得到的代数系统 ,目的是为智能系统中的条件推理建立一个数学基础 .Multiple source information forms one of the key components of data fusion. Such information may emanate from various mechanical sensor sources such as radar or Doppler systems, or it may derive from human-based sources, such as via expert opinion expressed through natural language. In general, each unit information is associated with degree of uncertainty/certainty which is traditionally determined through the use of probability. Thus, one can evaluate probabilistically any desired logical combination of events for use in decision-making. On the other hand, information uncertainty is provided in a way that there appears to be no single underlying Boolean event whose probability evaluation matches the prescribed uncertainty. Such uncertainty is often expressed in the form of given functions of probability evaluations of contributing simpler events. For example, when these functions are simple arithmetic divisions with arguments being pairs of events, each numerator argument event being a subevent of the denominator argument event, the quantitative uncertainty corresponding to each unit of information then, becomes a conditional probability. But, in general, the standard development of probability theory and statistics has not produced a way to represent conditional probabilities as single event probability evaluations, so that standard statistical decision-making techniques cannot be used in systematic sound way here. Recently, probability theory has been expanded to address the problem in the form of conditional event algebra. Conditional Event Algebra (CEA) is a relatively new logic system which rigorously extends standard probability theory to include events which are contingent such as rules and conditionals. The' if ...then' is modeled as Boolean elements, and yet compatible with conditional probability quantitative value. The principle theory and application of conditional event algebra is presented. We also introduce the idea of relational event algebra which is more general than conditional event algebra.

关 键 词：人工智能 知识工程 专家系统 条件事件代数 逻辑系统 模糊性推理 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]