因果知识的表示和推理  

The Representation and Reasoning of Causal Knowledge

作　　者：林海[1] 孙吉贵[1] 

机构地区：[1]吉林大学计算机科学与技术学院

出　　处：《计算机科学》2004年第5期123-126,共4页Computer Science

基　　金：国家自然科学基金(60073039;60273080);吉林省科技发展计划(20020306)

摘　　要：本文介绍了因果推理的两个主要应用:预测行为的间接结果,找出给定事实的真正原因,指出了用逻辑描述的因果关系在解决这两个问题中存在的不足。本文也简单地介绍了J.Pearl提出的因果推理方法,该方法的基本思想是把因果推理看成是一种“计算模式”,而这种把智能归结为“计算模式”的想法正是认知科学中的核心思想。本文最后比较了基于逻辑的方法和基于认知科学的方法实现智能的区别,指出了因果推理应该在认知科学的框架内得到解决。This paper presents two main applications of causal reasoning:to predict indirect facts of actions and to find the actual cause of some given facts. We point out that logic is not adequate for solving these two problems. We also present J. Pearl's method in dealing with causal reasoning,which treats causation as a computational schema. This idea of treating intelligence as computational schema is the basic idea of cognitive science. As last,we contrast methods which implement intelligence based on logic and those based cognitive science,point out that causal reasoning should be performed in the framework of cognitive science.

关 键 词：因果知识 因果推理 “计算模式” 逻辑 认知科学 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]