一种有效的Web用户访问模式聚类算法  

A Clustering Research for User Access Patterns in Web Logs

作　　者：郑富兰[1] 杨勇[1] 贺丽春[1] 

机构地区：[1]山西师范大学数学与计算机科学学院,山西临汾041004

出　　处：《山西师范大学学报（自然科学版）》2013年第1期39-42,共4页Journal of Shanxi Normal University(Natural Science Edition)

基　　金：山西省自然科学基金(2008011029-2);山西省大学生创新性实验项(2009)

摘　　要：用户在一个网站的访问行为反映了用户特定的兴趣,本文提出了一种用户访问模式的聚类算法.利用传统的Leader算法只扫描数据集一遍的优点,以及粗糙理论在处理含有不确定信息问题上的优势,给出了结合粗糙理论的改进Leader算法对用户存取模式进行聚类方法.通过一系列的实验分析,结果表明,该算法在可承受的计算时间内可对web日志中的用户存取模式进行有效聚类.聚类结果可应用于山西师范大学网站,对相应用户进行分析,给出个性化的服务等.Different sequences composed of visited web pages disclose users＇specific interest. Thus a cluste- ring ＂algorithm is proposed to cluster user access patterns in web logs. According to the advantages of traditional Leader algorithm, which only needs one data set scan, and that of rough set theory which can deal with issues having uncertainty information, a combination of improved Leader algorithm and rough set theory is proposed to cluster user access patterns, so as that there exists more similar behavior in one cluster and more evident differ- ence between any two clusters. The experiment result shows that an effective clustering can be done on the user access patterns in web logs at an acceptable computation expense.

关 键 词：粗糙集 用户存取模式 聚类分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]