基于概念格的规则产生集挖掘算法.pdfVIP

计算机研究与发展 Vo I.41 ,No.8 第 41 卷第8 期 2004 年8 月 JOURNALOFα)MPUTER RESEARCH AND DEVELOP岛fENT Aug. 2004 基于概念格的规则产生集挖掘算法 梁吉业1 ， 2 王俊红2 1( 中国科学院计算技术研究所智能信息处理重点实验室 北京 100080) 2( 山西大学计算机与信息技术学院太原 030006) (tj y@ sxu. edu. cn) 摘 要 传统的规则提取算法产生的规则集合相当庞大，其中包含许多冗余的规则.使用闭项集可以减少规则的数目， 而概念格结点间的泛化和例化关系非常适用于规则提取.基于概念格理论和闭项集的概念，提出了一种新的更有利于规 则提取的格结构，给出了相应的基于闭标记的渐进式构造算法和规则提取算法.最后提供给用户的是直观的、易理解的 规则子集，用户可以有选择地从中推导出其他的规则.实验表明该方法能够高效地挖掘规则产生集. 关键词 概念格;闭项集;规则产生集;规则提取 中图法分类号 TP18 An Algorithm for Extracting Rule- Generating Sets ßased on Concept Lattice 1 LIANG Ji_Ye ,2 and WANG Jun-Honi l(Key la如ratory of Intelligent In.如wωtion Processing , Institute of com如ting Technology ， αinese Acade1ηyof段位nc饵，应对ing 100080) 2( ，s油JOl of Computer and Information Technology ， 局anxi Uniærsity , Tai.沪剧 030006) Abstract The rule sets extracted by traditional algorithm 盯e usually very 1盯ge ， because it includes many redundant rules. The number of rules can be reduced usin:g closed item sets. The relationship of generalization and specialization 缸nong concepts of ∞ncept lattice is very suitable for extracting rules. A new and more advantageous lattice structure for extracting rules is pro归sed b臼ed on the theory of ∞ncept lattice and the concept of closed item set. Then , an incremental algorithm based on closed label for ∞nstructing lattice and algorithm for rules extracting are developed. Finally , a visual and easily understandable set of rules is presented to user , who can selectively derive other rules of interest. The example shows that the algorithm used in this paper can efficiently extract rule-generating set. Key WOI也 concept lattice;closed item set;rule-generating set; rule extracting 间的泛化和例化关系，而它的 Hasse 图则实现了对 同 吉