User Modeling Driven News Filtering Algorithm for Microblog Service in China

Sina microblog is an increasingly popular social network service in China. According to previous researches, microblog is considered to possess some characters of news media. However, except for the studies focusing on detecting real-time event spread around the globe, little work has been done to excavate news in daily life. In this paper, we conducted a study of filtering news in Sina microblog. We first generalized the definition of news by user surveys and quantitative analysis. Then we built a news recommendation system by modeling the users and applying several certified heuristic rules. Finally we evaluated the system and found out its satisfying performance in different datasets. By applying our news recommendation system, people can get news-worthy information for themselves while the entertaining tweets, which are omnipresent in microblog, are put in the last of the news list. Therefore we can achieve a much better understanding of heuristic rules about news by this work, which is helpful to certain official organizations as they need to detect and supervise the popular news among the public.