一种基于路径聚类的Web用户访问模式发现算法  

An Algorithm for Discovering Customer Access Model Based on Web Path Clustering

作　　者：马力[1] 焦李成[1] 刘国营[2] 

机构地区：[1]西安电子科技大学雷达信号处理国家重点实验室,西安710071 [2]西安邮电学院信息中心,西安710061

出　　处：《计算机科学》2004年第8期140-141,172,共3页Computer Science

基　　金：陕西省自然科学基金(2002F26)基金资助

摘　　要：本文在将用户对 Web 站点访问行为表示成用户访问事务基础上,给出有关兴趣度、相似度、聚类中心定义,提出基于 ISODATA 算法的路径聚类方法,并对实际网站 Web 日志文件进行实验,结果表明该方法不但能够发现群体用户访问模式,而且还得到较为合理的模式聚类个数。According to the analysis of Web site's log file,firstly,the behavior of the customers access to Web pages can be represented as the access transaction of the customers. The interest ,similarity and cluster center based on the customer transaction above are defined. Then ,the new method of path cluster based on the algorithm of ISODATA is given. Lastly experiment to practical Website's is made. The result of the experiment verifies that the method can not only discover the access model of the customer groups ,but also obtain the more reasonable number of the model clus- ter.

关 键 词：WEB 事务 ISODATA算法 用户访问模式 路径聚类 

分 类 号：TP393.01[自动化与计算机技术—计算机应用技术]