专家系统中不精确推理模型的比较

COMPARISON OF INEXACT REASONING MODELS IN EXPERT SYSTEMS

在基于规则的专家系统中,通常采用四种不精确推理模型,即Bayesian概率理论,CF理论,D-S证据理论和可能性理论。本文在讨论这四种模型的基本概念和性能的基础上,对这四种模型各自的优缺点进行了比较,这为在研制一个实用的专家系统时选择一个合适的不精确推理模型提供了一个良好的依据。

In rule-based expert systems, four inexact reasoning models commonly used are Bayesian probability theory, CF theory, D-S theory of evidence and Zaden's possibility theory. In this article, we compare the advantages and disadvantages of these four models after discussing their basic concepts and performances. This provides some guidelines for selecting an appropriate inexact reasoning model in developing a real-world expert system.