中文摘要：

利用微博关注关系和社交行为构建微博信任网络,通过引入基于信任的随机游走模型,结合用户间兴趣相似度,建立了微博粉丝推荐模型。为提高粉丝推荐系统的覆盖率,将用户间的社交行为引入信任的计算,实现了TopN推荐。利用KDD Cup 2012腾讯微博数据进行了实证研究。实验结果表明：在混合多种社交行为的信任网络中,推荐算法的整体性能最优;推荐长度对推荐结果影响较大,当长度为40时算法获得最好的推荐性能;与主流的推荐算法相比,改进后的基于信任的随机游走推荐模型在推荐准确率和覆盖率等多种评价指标上都取得了更好的结果。研究结论为微博粉丝推荐研究提供了新的方法,为微博网络社会化推荐提供了新的视角。

英文摘要：

Microblogging followers relationship and their social behavior microblogging trust network. By introducing a random walk model of are employed to build trust, we establish the recommendation model for microblogging fans, combined with their interests similarity. To improve the coverage of the fans recommender system, their social behavior is used to calculate the trust degree and to implement the TopN recommendations. Data from KDD Cup 2012 is utilized to conduct the research. The experimental results show that the overall performance of the recommendation algorithm in the trust network mixed with a variety of social behavior are the best. The length of recommendation greatly influence the result of the recommendation. Recommendation algorithm with the lengths of 40 achieves the best performance. Compared with the dominating recommendation algorithm, the improved model of random walk recommendation based on trust outperforms the state-of-the-art recommendation algorithms in terms of accuracy and coverage. The findings provide a new method for fans recommendation research and a new perspective for social recommendation in microblogging network.