基于随机游走算法的社会化标签的用户推荐

User recommendation in social tagging based on random walk algorithm

利用来自Delicious的数据集,结合内容相似度的挖掘和语义关系处理,对社会化标签系统的用户推荐的算法进行了研究。具体工作为:利用标签和书签的语义关系,定义用户的内容信息,从而计算内容相似度;建立内容相似度与社会网络的用户链接关系,通过可重启的随机游走算法(RWR)结合来达成理想的效果。实验评测显示,无论是精确度还是召回率,该算法的效果都要明显优于baseline的算法。

By combining the similarity of contents and the semantic relation, the algorithms of user recommendation in social tagging system are studied by using the dataset derived from delicious, which is the world＇s leading social bookmarking service. This work mainly includes two aspects. Firstly, the similarity of contents is calculated by using the semantic relation between the tags and bookmarks; Secondly, random walk with restarts （RWR） algorithm is used to combine the similarity of contents with user links in social networks to achieve the desired effect. Experiments conducted on real datasets from delicious demonstrate that this method outperforms other temporal models and state-of-the-art prediction methods both from Drecision and recall.