区间值决策信息系统的启发式属性约简  

LIANG Chunhua,ZHANG Haiyun.Heuristics attribute reduction in interval-valued decision system

作　　者：梁春华[1,2] 张海云[2] 

机构地区：[1]山西省财政与税务专科学校经济信息系,太原030024 [2]山西大学计算机与信息技术学院,太原030006

出　　处：《计算机工程与应用》2012年第27期114-118,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(No.71031006;No.70971080;No.60903110)

摘　　要：区间值决策信息系统是单值信息系统的一种推广,借助于属性区间值的相似程度在区间值决策系统上引入α极大相容类的概念,定义了一种新的条件信息熵,提出了相对属性内(外)重要度的度量方法,进一步,给出基于α条件信息熵的启发式相对约简算法,通过实验验证了该算法的有效性。Interval-valued decision information systems are generalized models of single-valued information sys- tems. A kind of a maximal tolerance class is introduced by similarity grade of attribute＇ s interval-value in inter- val-valued decision system. This paper defines new conditional entropy among attributes in interval-valued informa- tion systems and proposes two types of measurement of relative attribute importance, which is inner attribute impor- tance and outer attribute importance. Furthermore, a heuristic relative attribute reduction algorithm based on a condi- tional information entropy is given, and the validity of the algorithm is illustrated by some experiments.

关 键 词：区间值决策系统 α极大相容类 α近似约简 相对属性重要度 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]