一种广义信念修正的计算方法  

A Computational Method of General Belief Revision

作　　者：贾松茂[1] 高阳[1] 孙召春[1] 陈世福[1] 

机构地区：[1]南京大学计算机软件新技术国家重点实验室,南京210093

出　　处：《计算机研究与发展》2004年第5期769-773,共5页Journal of Computer Research and Development

基　　金：国家自然科学基金项目 (60 10 3 0 12 );国家"九七三"国家重点基础研究发展规划项目(2 0 0 2CB3 12 0 0 2);江苏省自然科学基金项目(BK2 0 0 3 40 9)

摘　　要：AGM理论中新知识为单个语句 ,而实际上新知识通常是以多语句甚至无穷语句的形式出现 广义信念修正从理论上扩充了AGM系统 ,但在具体实现上还缺乏有效的算法 通过构造优序划分的方法解决信念间的关系问题 ,采用新知识的信度与冲突知识相关的方法解决新知识的信度问题 给出了一种基于链表的计算方法 。New knowledge in the theory of basic AGM postulates is a single sentence, but in fact it comes in set of sentences, even infinite sentences General belief revision extends the AGM's theory, but a valid computational method is wanted The problem of the relation between believes can be resolved through developing the nice ordering partition model, and the problem of how to determine the new knowledge's epistemic entrenchment can be resolved through the method of correlating the new knowledge with the conflict knowledge A computational method based on linked list is given, which resolves the operating problem about the belief library of general belief revision

关 键 词：信念修正 广义信念修正 AGM GBR 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]