一种基于浓缩树结构的属性约简算法  

AN ATTRIBUTE REDUCTION ALGORITHM BASED ON CONDENSED TREE STRUCTURE

作　　者：张忠平[1] 林志杰[1] 曹丽君[2] 高宝江 裴玉芳[4] 

机构地区：[1]燕山大学信息科学与工程学院,河北秦皇岛066004 [2]河北科技师范学院,河北秦皇岛066004 [3]中国工商银行秦皇岛分行,河北秦皇岛066004 [4]秦皇岛市第一医院,河北秦皇岛066004

出　　处：《计算机应用与软件》2011年第8期76-79,共4页Computer Applications and Software

基　　金：国家自然科学基金(60773100);河北省教育厅科研计划项目(2006143)

摘　　要：属性约简是粗糙集理论的重要研究内容之一,以浓缩树结构存储区分矩阵中的非空数据项,对IReductBtree算法进行了一些改进。它是根据树结构特点结合特殊的启发式策略,删除树中结点,直到树空最后得到信息系统的一个约简。最后,通过一个实例完整演示了该方法证实其有效性。Attribute reduction is an important area to study in rough set theory.This paper suggests storing occupied data items in the differentiated matrix with condensed tree structure storage to achieve some improvements on IReductBtree algorithm.It employs condensed tree features with a special heuristic strategy to gradually delete nodes from a tree until it becomes null in the end.Then a reduction of the information system is generated.An instant in the end demonstrates the entire process of the improved method and verifies its validity.

关 键 词：粗糙集 区分矩阵 属性约简 浓缩树 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]