摘要： 提出一种基于Web日志的分析方法，构建UserID-URL关联矩阵，通过矩阵聚类进行页面预测，为了加快聚类速度和效果，引进Ping-Pong算法，实验表明基于矩阵聚类的页面预测达到较高的命中率。
关键词:
矩阵聚类,
Ping-Pong算法,
页面预测
Abstract: In this paper, a new method of analysis on Web log is proposed. A UserID-URL matrix is created and it achieves Web page predicting through matrix cluster, additionally a Ping-Pong algorithm is introduced in order to increase the speed of cluster. The experiments show that the Web page predicting through matrix cluster has high hit probability.
中图分类号:
彭玉青,田海山,陈美娜. 基于矩阵聚类的页面预测研究[J]. , 2004, 30(08): 161-163.
PENG Yuqing, TIAN Haishan, CHEN Meina. Research of Web Page Predicting Based on Matrix Cluster[J]. , 2004, 30(08): 161-163.