集值决策信息系统的属性约简及确定性决策规则优化  

Attribute reduction and definite decision rules optimization in set-valued decision information systems

作　　者：管延勇[1] 薛佩军[1] 胡海清[1] 

机构地区：[1]山东大学数学与系统科学学院

出　　处：《系统工程与电子技术》2006年第4期551-555,619,共6页Systems Engineering and Electronics

基　　金：国家自然科学基金(70271048);山东省自然科学基金资助课题(Y2001G07)

摘　　要：在集值决策信息系统中定义了一种相容关系,利用最大相容类定义了两类上下近似算子;分别利用两类下近似算子,定义了属性的A-相对约简和E-相对约简,以获取确定性决策规则的优化规则;分别定义了两种区分函数,相应地给出了A-相对约简和E-相对约简的区分函数求法,并证明了该方法的正确性。In set-valued decision information systems, some objects may possess more than one attribute values for an attribute. A tolerance relation and two kinds of lower and upper approximation operators are defined by use of maximal tolerance classes, and two kinds of attribute reduction are derived accordingly, that is, A-relative reduction and E-relative reduction, so as to get the optimal definite decision roles. In order to compute A-relative reduction and E-relative reduction, two discernibility functions are defined respeetively, and the computation approaches of these two kinds of attribute reduction are proved and exemplified.

关 键 词：信息科学 信息系统 粗糙集 属性约简 区分函数 

分 类 号：TP120.99[自动化与计算机技术—控制理论与控制工程]