摘要
基于Web日志的信息挖掘具有重要的意义,比如识别兴趣相似的客户群体有利于实现推荐和个性化服务。采用了多元线性回归分析用户浏览行为,直接对兴趣相似矩阵进行λ截聚类,最后通过计算项与类的连接强度来调整聚类结果。实验结果证明了该算法具有较高的准确率和良好的扩展性。
Data mining based on Web logs is of great significance. For instance,it can discover groups of people with similar interests and facilitate recommendation and personal service. A new clustering method based on Web users’ interests regressively analyzes users’ behaviors,partitions the interesting matrix with a threshold λ,and finally relocates some elements of clusters based on the joint strength between an element and a cluster. The favorable precision and scalability of the algorithm are studied through the experiments.
出处
《计算机系统应用》
2010年第4期62-65,共4页
Computer Systems & Applications
基金
湖南省科技计划基金(2006JT1040)
