一种改进的正负关联规则挖掘算法  

Improved Positive and Negative Association Rules Mining Algorithm

作　　者：陈宁军[1,2] 高志年[1] 

机构地区：[1]南京陆军指挥学院作战实验中心,南京210045 [2]南京陆军指挥学院研究生三队,南京210045

出　　处：《计算机科学》2011年第12期191-193,212,共4页Computer Science

基　　金：中国博士后科学基金(200902695)资助

摘　　要：针对传统正负关联规则挖掘算法需要多次扫描数据库并且生成大量候选频繁项集的问题,在对比目前相关研究成果的基础上,提出了一种改进的正负关联规则挖掘算法,它通过两次数据扫描完成对正负关联规则的挖掘,对最大频繁项集的挖掘算法做了改进,有效提高了算法效率,同时对置信度标准做了改进。基于某真实事务集的实验表明,算法提高了规则挖掘的质量和有效性。Aiming at the problems of the traditional positive and negative association rules, such as multi-scanning data- base and generating large candidate frequent itemset,on the basis of comparing recent research on association rules mi- ning,put forward an improved positive and negative association rules mining algorithm, which can accomplish mining positive and negative association rules by scanning Dataset twice; and improve the algorithm of mining the maximal fre- quent itemset, which result in upgrading the efficiency of the algorithm; besides, improve the standard of confidence le- vel, in order to increase the quality of association rules mining.

关 键 词：正负关联规则 关联规则挖掘 最大频繁项集 置信差 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]