基于优势度的序信息系统属性重要性度量  

Measure of Attribute Importance Based on Dominant Degree for the Ordered Information System

作　　者：李敬[1] 王利东[1] 李晓庆[1] 

机构地区：[1]大连海事大学数学系,辽宁大连116026

出　　处：《琼州学院学报》2015年第5期17-21,共5页Journal of Qiongzhou University

基　　金：国家自然科学基金(61203283);辽宁省自然科学基金(2014025004);中央高校基本科研业务项目(3132014036;3132014324)

摘　　要：属性约简在某种指标下最大限度保留信息系统的主要信息,使所获得的知识简化,其有利于获取有效的决策规则.针对序信息系统约简问题,本文构建了两种基于优势度的属性重要性度量方法及探索其相关数学性质,并将其应用于企业发展能力信息系统获得属性约简集合.实例分析表明了所构建方法能有效地利用数据的分布信息来反映出对象之间的差别程度,更客观体现出属性的重要程度.Attribute reduction can preserve main information for the information system with respect to some indexes,from which some simplified knowledge can be obtained for decision makers. In order to address the attribute reduction in ordered information system,two methods of attribute reduction are proposed based on dominance degree. In addition,the mathematical properties of dominance degree and dominance sort entropy are studied in the current research. The proposed methods effectively reflect the degree of difference between objects by using dominance degree. An example is experimented on the information system of business development and shows that the proposed method is effective for attribute reduction.

关 键 词：序信息系统 优势度 属性约简 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] O236[自动化与计算机技术—控制科学与工程]