一种基于信息熵的信息系统属性约简算法  

An Algorithm of Attribute Reduction of Information System Based on Entropy

作　　者：吕林霞[1] 赵锡英[1] 唐占红[1] 

机构地区：[1]兰州工业学院软件工程学院,甘肃兰州730050

出　　处：《自动化与仪器仪表》2013年第5期197-199,共3页Automation & Instrumentation

基　　金：甘肃省自然科学研究基金计划资助项目(1208RJZA186);甘肃省技术研究与开发专项计划资助项目(1205TCYA037)

摘　　要：粗糙集中找到最小属性约简是一个NP-hard问题,本文根据知识信息熵的定义和性质,定义属性重要性,分析其性质,得出核的求法,给出约简的判定定理,提出了一种基于信息熵的属性约简算法。该算法采用启发式搜索法,先计算属性重要性,求得核,再以核为起点,以属性重要性大小为启发式信息,选择属性,求得最小约简。理论分析和实际计算表明,该算法简明有效。Finding minimum attributes reduction is a NP-hard problem in rough set. This paper defined importance of attributes based on definition and properties of knowledge entropy, made the analysis of its nature, gave the solution of core and the reduction theorem and proposed an algorithm of attribute reduction based on entropy. The algorithm uses heuristic search method. First it cal- culates the attribute importance and gets core, then taking the core as a starting point and making size of attribute importance as heu- ristic information, it selects attributes and gets the minimum attributes reduction. The theoretical analysis and example calculation show the algorithm is simple and effective.

关 键 词：粗糙集 信息熵 属性约简 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]