一种基于生成树的分类去除关联规则冗余方法  

A Method of Removing Redundant Association Rules by Classification Based on Spanning Tree

作　　者：楼巍[1] 邓柳燕[1] 严利民[2] 郭丽媛[2] 

机构地区：[1]上海大学机电工程与自动化学院,上海200072 [2]上海大学微电子研究与开发中心,上海200072

出　　处：《计算机技术与发展》2014年第6期24-27,31,共5页Computer Technology and Development

基　　金：教育部人文社会科学研究专项课题(10JDSZ1015);上海学校德育创新发展课题(2012-dycx-105)

摘　　要：在信息及数据爆炸的时代,冗余问题已经成为数据挖掘者获得知识的重大障碍,而目前解决的方法会导致关联规则的不完整性。基于此,文中引入了有向超图表示关联规则,重定义了邻接矩阵,介绍了冗余规则分类处理思想,将冗余规则分为从属规则和重复路径规则,通过VB编程去除了从属规则冗余,以及利用生成树算法去除了重复路径规则冗余。实验结果证明,此方法创新性地结合了图论中有向超图、生成树与关联规则的知识,维护了关联规则的完整性和准确性,同时去除了全部冗余规则。In the information and data explosion era,redundant rules problem has become a major obstacle to gain knowledge for data miners,but the current solution may lead to the incompleteness of association rules.For the issues above,introduce the directed hypergraph to express association rules,redefine the adjacency matrix,propose the categorical thinking for removing redundant rules,dividing redundant rules into slave rules and repeated-path-rules,the algorithm of removing the dependency rule by VB programming and the spanning tree algorithm to remove the redundancy of repeated-path-rules.The result of experiments show that this method is effective,fast and to maintain the integrity and accuracy of association rules which links directed hypergraph,spanning tree of graph theory and association rules innovatively.At the same time it removes all redundant rules.

关 键 词：关联规则 有向超图 邻接矩阵 生成树 冗余 去除从属规则 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]