Personalized Search Of Weibo Concerned About User-Publisher Relationship

With the rapid development of the WEB 2.0, the social network applications such as Wechat, Weibo are widely used. While the massive and real-time information brings user fun, searching what they actually interested in becomes increasingly difficult. Compared with the traditional web personalized search, Weibo and the others have advantages in strong sociality, fruitful topics and shorter text, which the method of traditional web personalized search becomes unavailable. Otherwise, every Weibo has its specific publisher, which makes the users concern not only each Weibo but also the publisher who writes it when they searching in the social network. Therefore, the theoretical study about how to take account of Weibo content, the information of users themselves, and enhance the personalized search effect should be further investigated.In this paper, it puts forward a topic model and language model to the users. It integrates the users in interested information of providers, including the similarity on interested topic model, the people of mutual followers, and the social network text information which in the length of context, link, hashtag and repost. The model also concerns about the relationship of user and publisher, the characteristic of publisher and the characteristic of Weibo. This paper takes above three angles to model the user. Finally, this paper uses the web spider to crawl data from Sina Weibo and the real user feedback to search to verify the effectiveness of user model for experiment.On the basis of modeling, this paper designs a set of personalized sorting mechanism in ranking method merges the user-publisher relationship, publisher character and twitter quality to get user final score and rank. Web crawler was used to get Sina Weibo data, real user data was used to verify the model with experiment. We designed a personalized search framework concerned about user publisher association. The results showed that considering the author information can improve search performance effectively and provide users better personalized search of twitter.