Interest Topics Oriented Approach For Sina Weibo Personalized Recommendation

As a typical social network media,micro-blog attracts a large number of users through its own unique features.With the increasing number of users,micro-blog's hot topics are constantly emerging.More and more researchers and scholars are keen to explore the useful information contained in micro-blog data,so as to tap the characteristics of users' favorite micro-blog.However,because of the complex content of micro-blog,when facing massive micro-blog text,users usually can not quickly and accurately find their own interested micro-blog,and the user's needs are personalized and diversified.Therefore,in the massive micro-blog information,how to find the micro-blog information of interest for users is an important problem to be solved urgently.Around this issue,the main work of this paper is as follows:(1)user micro-blog interest topics mining: Based on the micro-blog information of users' history,the micro-blog information of each user is modeled as a document,and the probability distribution matrix of the user's micro-blog vocabulary is obtained by the LDA(Latent Dirichlet Allocation)topic model,and then mining the interesting micro-blog topics from the user,and recommending personalized micro-blog.Lay the foundation.(2)personalized recommendation of micro-blog based on topics similarity and forwarding amount: in order to recommend high quality hot micro-blog to users,it is considered that the traditional micro-blog recommendation only uses the similarity between the subjects as the recommendation score,and does not make use of the functions of the micro-blog itself(such as the forwarding amount,the number of comments,the number of fans,etc.).Therefore,this paper calculates the similarity between topics of user interest and the current hot micro-blog topics,and considers the number(forwarding amount)of the user forwarding micro-blog,combining topics similarity with the micro-blog forwarding amount to calculate the micro-blog's recommendation index,and recommends the hot spot of the high index to the user.Finally,using Sina micro-blog test dataset as an experimental carrier,the effectiveness of the proposed method is verified by experiments.