Research On Personalization Recommendation Method Based On Micro-Blog Users’ Interest

With the recent rise of fiery web3.0, the development of the social network is very hot. The rapid development of the social network such as RenRen, QZone, Microblog, Facebook have attracted large number of users. In recent years, in terms of domestic social network, the microblog is very fascinating. More and more people know the news nearby, society, the word by the real-time information on microblog platform. But with the increasing number of the microblog users, there is billions of microblog information every day. The microblog users face the serious phenomenon of "information overload". How to let a microblog user get away from the phenomenon and find the information that interested in the microblog through the users that interested in is a problem worthy of studying. This study is very important for the improvement of users’ usage experience and the quick access to information.In this paper, we research an important aspect of social network-Microblog, and we mainly analyze and study the Personalized recommendation problem on Microblog users. Research and design the user personalized recommendation algorithm, and according to the actual demand, designed and completed the microblog user personalized recommendation system.The main research Achievement about this paper includes:1)This paper proposes a microblog-content expand algorithm. It can enhance the semantic information of microblog content and improve the effect of topic model’s building by the correlation between microblog operations. Experiment shows that the algorithm has an obvious improvement on the building of topic model.2)This paper proposes a strategy about the selection of microblog friends candidates. It can determine the scope of the suggested users by the feature of microblog followee/follower network.3)This paper proposes the method of building different types of user interest model. Using LDA model compute different users’interest, based on this, the personalization recommendation method for microblog users is proposed. Experiment shows that this method is more accurate than the method based on TF-IDF.4)This paper designs and implements a microblog user recommendation system. This paper will introduce the overall design of the system from functional partitioning, hierarchical design and database design. The implement process of this system will be introduced briefly in this paper on different functional modules including login information authentication, personal information browsing, recommended users browsing, user information management and log information management. The system can recommend the microblog users they interested in, and let the users get rid of the problem of information overload.