Real-time and personalized recommendation on microblogging systems

Social network services need to response to the real-time, continuous and personalized requirements. However, most social network services do not personalize user requirements. Large volume updates of user generated content result in a challengeing task that a social network system provides a real-time, continuous and personalized service. To better response to the real-time, continuous and personalized requirements, in this paper, we infer the topic distribution of microblogs and interest vector of a user based on the LDA model. Based on the topic model, we further recommend user interested microblogs which are published recently on microblogging systems. Finally, we have conducted extensive experiments on the real dataset, and elaborated the effectiveness and efficiency of our proposed method.