一种新的不一致决策表属性约简算法  

New algorithm for attribute reduction in inconsistent decision tables

作　　者：汪小燕[1] 杨思春[1] 

机构地区：[1]安徽工业大学计算机学院,安徽马鞍山243002

出　　处：《计算机应用》2008年第2期525-527,共3页journal of Computer Applications

基　　金：安徽省教育厅自然科学基金资助项目(KJ2007B245)

摘　　要：针对目前求核方法存在的问题,提出一种基于分布函数的用于计算核属性的改进的二进制可辨矩阵。改进的二进制可辨矩阵不仅规模小,而且适用于任何决策表求核。在获取核属性的基础上,提出一种新的不一致决策表的属性约简算法,只要在用于计算核属性的改进的二进制可辨矩阵中简单增加相应的行,就可以利用逻辑运算来获取属性约简。并将吸收律应用于属性约简,较大提高了属性约简的效率。In order to correct some problems of computing the core of a decision table, an improved binary discernable matrix definition based on distribution function for computing the core was put forward in this paper. The improved binary discernable matrix is not only small in scale but also suitable for any decision tables for computing the core. A new algorithm for attribute reduction in inconsistent decision tables was also presented based on getting the core. Only increasing appropriate rows in the improved binary discernable matrix for computing the core, attribution reduction will be obtained by utilizing logic operation. And the absorptive law was used in attribution reduction, which raised the efficiency of attribute reduction greatly.

关 键 词：粗糙集 不一致决策表 二进制可辨矩阵 核 属性约简 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]