网络书签系统中基于社团结构的个性化推荐方法

An Approach for Personalized Recommendation Based on Community Structures in the Social Bookmarks System

面对互联网中大量共享的网络书签资源,单个Web用户显然无法充分挖掘它们的潜在价值,迫切需要在网络书签系统中嵌入个性化服务工具。考虑到Web用户在网络书签系统中收藏的Web资源能够有效反映其兴趣,本文提出了一种网络书签系统中基于社团结构的个性化推荐方法。该方法首先依据Web用户收藏的Web资源之间的重叠关系建立Web用户关系网络,然后利用CPM算法划分Web用户关系网络中的社团结构,最后利用这种社团结构实现社团内基于协作过滤的个性化推荐和社团间基于＂信息桥＂的个性化推荐。实验结果表明了该方法的有效性。

Facing the vastly shared social bookmarks resources,the single Web user cannot utilize the latent value of them efficiently,which calls for personalized services in the social bookmarks system.Considering that social bookmarks system collects the favorite Web resources of Web users,so this article proposes an approach of the personalized recommendation based on community structures in the social bookmarks system.Firstly,a social network of Web users is constructed based on the overlapping relation of the Web resources which are bookmarked by Web users.Then the social network is divided into several communities based on CPM.Finally,depending on the community structures,the personalized recommendation can be produced based on collaborative filtering in a community,or based on＂information bridges＂between communities.The experimental results show that the approach is efficient.