基于改进的AprioriAll算法的Web序列模式挖掘研究  

Research of web sequential pattern mining based on improved AprioriAll algorithm

作　　者：吴海燕[1] 朱靖君[1] 高国柱[1] 程志锐[1] 

机构地区：[1]清华大学计算机与信息管理中心,北京100084

出　　处：《计算机工程与设计》2010年第5期921-924,1034,共5页Computer Engineering and Design

基　　金：国家863高技术研究发展计划基金项目(2007AA010306)

摘　　要：为了减少AprioriAll算法挖掘过程中候选序列的生成以及对序列数据库的扫描次数,提高算法的挖掘效率,提出了一种基于改进的AprioriAll算法的Web序列模式挖掘方法。首先对数据进行预处理,然后利用经过改进的AprioriAll算法进行模式挖掘。算法的改进主要有两点:一个通过改变候选序列的连接方式来减少候选序列的产生;二是通过减少不必要的数据库扫描操作来提高算法的效率。通过实验验证了改进后算法在Web序列模式挖掘过程中的高效性和正确性。To reduce the generation of candidate sequences and the scans to sequence database for AprioriAll algorithm, an efficient sequential pattern mining method based on improved AprioriAll algorithm is presented. Firstly, data are preprocessed. Then the sequential pattem mining is finished by improved AprioriAll algorithm. The improvements of AprioriAll algorithm are mainly two points： one is to change the connection of candidate sequences to reduce the generation of candidate sequences; the other is to reduce the needless database scans to improve the efficiency of algorithm. Finally, the efficiency and validity of improved AprioriAll algorithm is validated by experiments.

关 键 词：AprioriAll算法 序列模式 WEB日志挖掘 事务 最大向前路径 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]