首页 | 本学科首页   官方微博 | 高级检索  
     

面向对象修改的核属性更新算法
引用本文:葛浩,李龙澍,杨传健,王波.面向对象修改的核属性更新算法[J].四川大学学报(工程科学版),2012,44(3):118-126.
作者姓名:葛浩  李龙澍  杨传健  王波
作者单位:1. 滁州学院机械与电子工程学院,安徽滁州,239012
2. 安徽大学计算机科学与技术学院,安徽合肥,230039
3. 滁州学院计算机与信息工程学院,安徽滁州,239012
摘    要:决策表中的对象被修改,核属性将会发生变化,需要对核属性及时更新。为此,首先,构造决策表等价类链表存储结构,并引入基于该存储结构的简化决策表定义和基于简化决策表的核属性定义;同时证明了该核属性与正区域核属性是等价的。然后,讨论修改指定对象情况下核属性更新方法,并给出相应的算法实现;算法以简化决策表为基础,仅需对等价类链部分搜寻,查找待修改对象,缩小了查找范围,提升了算法效率;另外,采用核属性计数方法,避免可分辨矩阵的存储,降低了空间复杂度。最后,通过实例验证了该算法的有效性。

关 键 词:粗糙集  正区域  可分辨矩阵  核属性  更新算法
收稿时间:2011/12/16 0:00:00
修稿时间:2012/2/20 0:00:00

Updating Algorithms of Core Attribute Based on Modifying Data
Ge Hao,Li Longshu,Yang Chuanjian and Wang Bo.Updating Algorithms of Core Attribute Based on Modifying Data[J].Journal of Sichuan University (Engineering Science Edition),2012,44(3):118-126.
Authors:Ge Hao  Li Longshu  Yang Chuanjian and Wang Bo
Affiliation:School of Mechanical and Electronic Eng., Chuzhou Univ.;School of Computer Sci. and Technol., Anhui Univ.;School of Computer and Info. Eng., Chuzhou Univ.;School of Mechanical and Electronic Eng., Chuzhou Univ.
Abstract:It is necessary that the core attributes of decision table will be updated dynamically, when the decision table object is modified. In order to improve the efficiency of computing core attribute and be beneficial to the large data sets, the storage structure of equivalence class list table of decision table is constructed, and the definitions of the simplified decision table and core attributes based on the simplified decision table are proposed. It is proved that the core attributes acquired from the definition is equivalence to core attributes of positive region. And then, the theories and algorithms of updating core attribute are researched and described on the aspects of modified the object appointed. Finally, the validity of the algorithms is verified in instances.
Keywords:rough set  positive region  discerniblity matrix  core attribute  updating algorithm
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《四川大学学报(工程科学版)》浏览原始摘要信息
点击此处可从《四川大学学报(工程科学版)》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号