摘要
属性约简是信息系统中的一个重要操作,而分类是属性约简的基础,且直接在大数据集上进行属性约简往往存在效率低下的问题。故以分类为基础提出了一种基于信息熵的信息系统并行属性约简算法。该算法通过信息熵的计算,在属性约简的同时对原信息系统逐层分解成尽量均匀的子表,从而实现了属性约简的并行计算并缩小了搜索空间。对该算法的时间复杂度进行了分析,实验表明,该算法在效率方面优于传统算法。
The reduction of attributes is an important operation of information system,and classification is the basis of attribute reduction.The attribute reduction would generally be inefficient if the huge data sets were mined directly by the algorithm.So this paper proposed a parallel algorithm of attributes reduction of information system which based on entropy of information.Through calculating the entropy of information,this algorithm which on basis of classification disassembled the information system into multi-table as symmetrical as possible while reducing the attributes.So the algorithm achieved a parallel calculation of attribute reduction and lessen the searching roomage.This paper also analyzed the time complexity of the algorithm.The experiment shows that the new algorithm is more effective than the traditional algorithm.
出处
《计算机应用研究》
CSCD
北大核心
2010年第11期4124-4127,共4页
Application Research of Computers
基金
甘肃省科技支撑计划资助项目(1011GKCA040)
兰州市企业技术攻关计划资助项目(2009-01-04)
兰州交通大学"青蓝"人才工程基金资助项目(QL-05-10A)
关键词
信息系统
属性约简
信息熵
并行算法
information system
attributes reduction
entropy of information
parallel algorithm
