融合用户兴趣与信任的微博推荐

Microblog Recommendation Based on Interests and Trust

在微博社交网络中,微博用户每天针对热门新闻、事件等生成众多微博内容,导致用户在大量内容中找到自己真正感兴趣的信息非常困难。因此,系统向用户推荐其感兴趣的微博,是改善用户体验的重要途径。提出一种新的模型因子分解机FM,以及综合考虑用户兴趣与信任因素的预测方法ITFM,以提高个性化微博推荐质量。通过在真实的数据集上进行模拟实验,结果表明,所提出的微博推荐方法在一定程度上提高了微博推荐准确度。ITFM方法能够有效解决信息过载问题,对改善用户体验具有较好的理论和实际意义。

Microblog users generate numerous microblog contents based on breaking news and latest events every day.However,it is difficult to find information of interest from these contents.Recommending interesting microblogs from the Microblog system is an important way to improve user experience.In this light,we build a model called ITFM,which combines factorization machines together with user interests and trust factors to improve the quality of personalized microblogging recommendations.Through simulations on real data sets,results show that the proposed Microblog recommendation approach improves the accuracy to some extent.ITFM can effectively deal with the information overload problem,and our work has better theoretical and practical significance for improving user experience.