专家知识库粗集建模中基于熵的数据离散化  

Dicretization based on Entropy in Modeling of Expert Knowledge Base with Rough Set

作　　者：包振强[1] 王宁生[1] 李斌[2] 

机构地区：[1]南京航空航天大学CIMS研究中心,南京210016 [2]扬州大学信息学院,江苏225009

出　　处：《数学的实践与认识》2003年第8期60-65,共6页Mathematics in Practice and Theory

基　　金：国家 8 63 /CIMS项目 (863 -5 11-910 -4 0 3 );江苏省教育厅自然科学基金重点项目 (0 1KJB5 2 0 0 10 )

摘　　要：首先分析了专家知识库粗集建模中连续数据离散化存在的问题 ,指出允许引入少量的冲突对专家知识库的建模分析是有益的 ;提出了一种基于信息熵的数据离散化方法 ,并分析了数据离散化的熵的度量 ,根据求解问题设计了一种问题求解的遗传算法 ;最后以基于多 Agent车间调度系统中调度The paper analyze the problems of th e real-value disretication in a rough set model for a expert knowledge base and point out that the permission to introduce a few data conflicts in the process o f modeling is helpful to the analysis of expert knowledge at first. A method for dicretization of attribute values based on the information entropy is presented , in which we measure the dicretization with the entropy gain of an attribute an d we construct a genetic algorithm for the method. Finally we give an example of the application of the method that is the disretication in modeling of the know ledge base of scheduler agents.

关 键 词：专家知识库 粗集 熵 数据离散化 冲突 阈值优化 遗传算法 应用 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]