基于粗糙集信息观的决策表属性约简方法  

Method of attribute reduction of decision table based on the information view of rough sets theory

作　　者：朱六兵[1] 唐德波[1] 杨斌[1] 

机构地区：[1]西南交通大学计算机与通信工程学院,成都610031

出　　处：《信息技术》2006年第1期46-49,共4页Information Technology

摘　　要：粗糙集理论是近年来发展起来的一种有效的处理不精确、不确定、含糊信息的数学理论方法,它被广泛应用于相容和不相容决策表的属性约简和核属性计算。利用反例指出目前基于粗糙集信息观[2、6]的决策表属性约简和核属性计算方法的局限性。对决策表的性质作了深入的研究,研究发现文献[2、6]方法的不足原因是:它们没有考虑U/ind(C)中等价类的相容性。给出了基于U/ind(C)中等价类相容性的属性约简定义和核属性定义,并给出了一种新的基于粗糙集信息观的决策表属性约简和核属性计算方法。讨论了该方法同文献[2、6]方法的区别。最后用相同实例验证了该方法的有效性。Rough set is a valid mathematical theory developed in the recent years, which has the ability to deal with imprecise, uncertain, and vague information. It is widely applied for attribute reduction and core attributes computation of consistent and inconsistent decision table. With a counterexample the paper points out an error of the method of the computation of attribute reduction and core attributes based on the information view of rough sets theory in the literature 2 and 6. In this paper, the properties of decision table are studied in depth, which shows that the error reason of methods in the literature 2 and 6 is that they didn＇ t consider consistency of equivalent classes in the U/ind （ C）. Tile definitions of attribute reduction and core attributes are given based on the consistency of equivalent classes in the U/ind（C）. A new method of the computation of attribute reduction and core attributes is presented based on the information view of rough sets theory. The differences between methods in this paper and in the literature 2 and 6 are analyzed. In the end, the same example shows that the method in this paper is valid.

关 键 词：粗糙集 决策表 属性约简 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]