【摘要】 Web用户聚类是通过分析用户会话,将具有相同或相似访问特征的用户聚为一类。在会话相似性度量方面综合考虑了网页浏览时间和访问频次两个因素,并考虑到用户个人习惯、能力等因素对浏览时间的影响,将浏览时间处理为RDP(Reduce the Differences in Personality)浏览时间,以降低其个性特征。为此,提出一种基于用户特性的RDPk-means聚类算法。实验表明,该算法可以有效实现用户会话的聚类,聚类结果客观合理。更多还原

【Abstract】 The Web users’ clustering is to group the users with same or similar surfing behaviour into one class by analysing their sessions.In this paper,two factors of browsing time on webpage and visiting frequency are synthetically considered in sessions’ similarity metric.In addition,the influence of other factors such as personal habit and ability on browsing time has also been taken into account.Browsing time is processed as RDP browsing time so as to reduce its personality characteristics.Therefore,we propose a personality characteristicsbased RDPk-means clustering algorithm.Experiments show that this algorithm is effective in realising users’ sessions clustering,the clustering results are objective and reasonable.更多还原