摘要
针对当前推荐系统研究中存在的问题,提出了一个基于Web日志的Internet协作推荐系统.它在用户浏览兴趣度量时综合考虑了页面访问次数、浏览时间的长度和页面的大小,然后将具有相似用户浏览兴趣的页面进行推荐.算法过程如下:先对Web日志进行预处理后得到用户事务,然后对这些用户事务进行各个页面的浏览兴趣矢量表示,最后通过推荐引擎生成实时推荐.经实验表明,该系统比其他方法推荐的网页点击率高.
Considering the drawbacks of the present Internet recommendation systems, a Web log based collaborative recommendation system is presented. User navigation interest is represented by the navigation times, time, and the size of the page. The process of the recommendation system is that user sessions are firstly set up after preprocessing the Web logs and then presented by the vector of user navigation interest, and finally the real-time recommendation is supplied to users by recommendation engine. This system is experimentally proved to have a better accuracy and a higher click rate than other methods.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2002年第12期1271-1274,共4页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(80173058).
