基于关系积的属性约简算法  

Attribute reduction algorithm based on attribute union

作　　者：焦吉成[1] 高学东[1] 邓君堂[2] 鄂旭[3] 

机构地区：[1]北京科技大学经济与管理学院,北京100083 [2]济南钢铁集团总公司技术中心,济南250101 [3]辽宁工学院计算机系,锦州121001

出　　处：《北京科技大学学报》2008年第6期694-697,共4页Journal of University of Science and Technology Beijing

基　　金：中国博士后科学基金资助项目(No.2005038319)

摘　　要：粗糙集的属性约简是一个NP难问题,目前尚无高效的算法.基于集合理论,提出了关系积概念和基于关系积的属性约简算法,把决策表的属性约简过程转化为关系积的运算,减小了对决策表的扫描次数,提高了属性约简的效率;算法采用自底向上和宽度优先的搜索策略,可确保找到最小属性约简集.结合实例,给出了算法的具体实现.Attribute reduction of rough sets is an NP hard problem, but there is not a popular efficient algorithm presently. The attribute union concept based on the set theory and the attribute reduced algorithm based on this concept were presented. The algorithm translates the attribute reduction to find the attribute union, reducing the number of scanning the decision table and improving the reduced efficiency. The scanning strategy from bottom to top and with width priority can insure to find the minimal reduction. Also, an example was presented to describe the algorithm.

关 键 词：约简算法 关系积 属性 集合理论 粗糙集 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]