摘要
互联网迅速发展,个性化信息服务成为研究的热点之一。RSS标准提供了结构化的信息模式,便于信息搜索和概要浏览。该文针对基于RSS标准的新闻源,根据用户点击等隐式信息,通过文本相似判定,自动聚类形成用户兴趣子类。用户模型节点、信息类别和用户兴趣子类构成了三层结构的树状用户兴趣模型。信息类别与用户兴趣子类均有对应的兴趣度。用户模型的更新是通过用户兴趣子类的更新与相关兴趣度的更新完成的。通过此模型进行信息推荐还要保证适当的信息冗余度。该模型的个性化程度高且更新效果好。
With the quick development of Internet, Personalized Information Service has become one of the hot spots. Information based on RSS standard has structural pattern, which is easily searched and browsed. In this paper, we use the news source based on RSS and cluster all the news into several User Interest Classes(UIC) according to user's clicks or other implicit information, by using the judgment of text comparability. User model point, Information Sort (IS) and UIC form a user interest model with tree structure which has three layers. IS and UIC all have interest degrees. The updating of user model includes the updating of UIC and the updating of interest degrees. It is necessary to remain some additional information when using this model to recommend news to users. This User Model is well personalized and efficiently updated.
出处
《计算机仿真》
CSCD
2005年第12期45-48,共4页
Computer Simulation
基金
863重大专项3TNET宽带信息网流媒体业务分发集成平台的研制(2003AA103810)