摘要
定义了属性和属性集的区分矩阵以及区分矩阵的运算,表明了属性集的区分矩阵与论域按属性集的分类是1-1对应的,因此,区分矩阵代表了属性集的分类。以区分矩阵作为属性约简的标准,以属性对分类的贡献率作为贪心选择的次序,建立了一种针对无目标信息系统的属性约简以及增量式属性约简方法。该算法降低了处理增量式数据的复杂度,并且只涉及到矩阵的加法、非负矩阵转化为0-1矩阵、判断矩阵相等的运算,便于用计算机语言编程计算。
The discernibility matrix of attributes sets and its operations are proposed. It shows that discernibility matrix of attributes sets can represent the classification of objects based on attributes sets. Discernibility matrix can be the standard of attributes reduction. Contributing rates is used as the order of greedy choosing. Then algorithms of attributes reduction and incremental attributes reduction of information system are produced. The complexity of algorithm is simple, and the computer program of the algorithm is very easy to make.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第18期4293-4295,4298,共4页
Computer Engineering and Design
基金
四川省教育厅青年基金项目(07ZB043)
关键词
粗糙集
贪心算法
区分矩阵
增量式属性约简
信息系统
rough sets
greedy algorithm
discernibility matrix
incremental attributes reductions
information system
