一种高效的属性核计算方法  

An Efficient Approach to Compute the Feature Core

作　　者：赵军[1] 王国胤[2] 吴中福[1] 唐宏[1] 李华[1] 廖晓锋[1] 

机构地区：[1]重庆大学计算机科学与工程学院,重庆400044 [2]重庆邮电学院计算机科学与技术研究所,重庆400065

出　　处：《小型微型计算机系统》2003年第11期1950-1953,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金 ( No.6980 3 0 14 )资助 ;攀登特别支持费 ;重庆市科委攻关基金资助

摘　　要：属性核的确定对解决粗集理论中的属性约简问题具有重要的意义 .但文献对属性核的求解方法鲜有讨论 ,基本上都是直接引用 Hu X.H.提出的基于分辨矩阵的方法 ,该方法具有较高的空间复杂度和时间复杂度 .本文基于决策系统的一致性 ,定义概念“关键属性”,并证明关键属性集合与属性核之间的等价性 .在此基础上 ,提出一种新的计算属性核的算法 ,该算法不需要建立分辨矩阵 ,其时间复杂度和空间复杂度显著地低于 Hu X.H.The determination of feature core may be significantly contributory to solve the problem of feature reduction in rough set theory. However there is rare discussion in literature on the ways of computing the feature core of a system. As to our best knowledge,basically only the approach based on discernible matrix proposed by Hu X.H. is directly adopted frequently. This algorithm is still very complex on the demand of auxiliary space and running time. In this paper,the notion of key feature is defined based on the consistency of a decision system,then the equivalence relation between the set of all key features and the feature core is discovered. Lastly a new efficient algorithm is put forward to find out the feature core. This algorithm,with no demand on the construction of discernible matrix,greatly outperforms the method proposed by Hu X.H. for its obviously much less time and space complexities than that of Hus.

关 键 词：粗集理论 属性核 系统一致性 关键属性 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]