关于“两种新的决策表属性约简概念”的注记  

Note on "Two New Reduction Definition of Decision Table"

作　　者：徐章艳[1] 宋威[2] 杨炳儒[2] 高静[2] 侯伟[2] 

机构地区：[1]广西师范大学计算机系,广西桂林541004 [2]北京科技大学信息工程学院,北京100083

出　　处：《小型微型计算机系统》2007年第9期1686-1689,共4页Journal of Chinese Computer Systems

摘　　要：出于应用目的,许多学者提出了各种不同的属性约简概念.蒋思宇等在分析了基于正区域的属性约简和基于信息熵属性约简概念的差别后,提出了两种新的属性约简概念.通过实例分析,指出两种新的属性约简概念的不足,其根源是度量平均决策强度和决策熵的两个公式不具有单调性.根据原作者的出发点,给出了度量决策强度和决策熵的两个新公式,分别证明了这两个公式都具有单调性,并由此给出了基于决策强度和决策熵的属性约简概念.For the sake of different applications, various definitions of attribute reduction have been proposed by different researchers. By analyzing the differences between attribute reduction based on positive region and that based on information entropy, Jiang et al proposed two new definitions on attribute reduction. By analyzing an example, we pointed out the disadvantages of these two definitions. The reasons caused the disadvantages lie in that the two formulas for measuring mean decision power and decision entropy are not monotonous. According to the basic ideas of Jiang et al, two new formulas for measuring decision power and decision entropy are proposed at first. Then, it is proved that the two proposed formulas are monotonous. Accordingly, definitions of attribute reduction based on decision power and decision entropy are proposed.

关 键 词：粗糙集 决策表 正区域 信息熵 约简 决策强度 决策熵 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]