中文摘要：

信息过载是影响微博等社会化媒体平台消费者持续使用行为的重要原因。协同过滤推荐能有效解决信息过载问题，但既有研究未能在推荐系统中整合用户创造内容和社会网络关系，社会网络关系体现出了消费者的偏好。针对微博的用户创造内容和社会网络两要素，本文从关键词层次人手，引入向量空间模型描述用户对关键词偏好，设计社会网络修订系数修订用户相似矩阵，实现社会网络关系驱动的协同过滤推荐模型。实验结果表明，较于基准协同过滤推荐方法，本文所提出的基于社会网络修订的协同过滤推荐能更准确并有效地实现个性化推荐。

英文摘要：

Information overloading is one of important factor which impact users＇ continual usage behavior in micro- blog and other social media websites. Collaborative filtering recommendation is an effective approach to solve the problem. However, prior research did not consider the influence of both UGC （user generated content） and social networks at the same time. Social networks represent users＇ preference just as UGC. According to the feature of both UGC and social networks, the paper brings the vector space model into the user＇ s rating model from the perspective of keywords. Method of calculating similar matrix between users is designed by combining the social networks structure characteristics. The collaborative filtering method driven by social networks relations has been realized. The experimental results have shown that the approach of collaborative filtering driven by social networks structural proposed in this paper outperforms standard ways.