一种基于K-均值的用户行为聚类算法

A User Action Clustering Algorithm Based on K-means

针对电子商务系统中大多采取用户评分或购买数据进行聚类,较少进一步分析用户行为的现状,提出一种根据用户浏览商品时序分析用户兴趣的方法 .在此基础上先用Canopy算法进行数据预处理后使用K-均值算法根据用户兴趣实现用户聚类.采用KDD CUP2000数据集中的用户点击流数据中的用户浏览记录对算法进行实验,实验结果表明算法有较好的聚类结果 .

This paper presents a method based on analyzingthe users＇ interesting while their browsing goods, aiming atfurther exploring the current situation of their behaviors,as the e ﹣ commerce conducts the clustering based on the user rating data or user purchase data,less doesthe further analysis of their behaviors. On this basis, the Canopy algorithm was used to preprocess the data,and then use K ﹣ means algorithm to do the user cluste-ring. And in the end,the user click stream data of Gazelle. com provided by the KDD Cup 2000 was used to vali-date the algorithm,the experimental results shows that the algorithm has better clustering results.