Microblog User Interests Mining Method Based On Multi-Source Data Fusion

With the rapid development of Internet, social networks widely recognized all over the world.In terms of domestic social networks, more and more people begin to Microblog release information, the site has become the mainstream of huge amounts of information release, the study of Microblog also from explicit interest tags to mining Microblog itself to the content of the potential themes. LDA(latent Dirichlet allocation) model is more popular in recent years, a kind of supervision and the theme of the model, there have been some studies by the LDA model on Twitter data set theme digging, but research on the theme of Chinese Microblog mining is not much. Based on the interaction between the user interest mining method and the user interest mining method based on mutual information can offset from the perspective of two different users interested in mining method based on Microblog content of defects.This article crawl sina_weibo user data for different levels of multi-source data fusion Microblog user interest modeling research. The main research results include the following aspects:First, combining with the traditional LDA model in this paper, a suitable for Chinese Microblog supervised interest subject mining model, based on the subject mining Microblog generation model CTM-LDA. The model effective use of the transcendental subject information, according to the user information and integration of user-generated content mining Microblog users interested in topic.Secondly, based on the interactive relationship and interaction of information sources such as building interest model respectively, based on the interaction between matrix and focus on the people interested in labels and the similarity between words generated interest focuses on people.Finally, in view of the microblogging custom content, topic, interactive information, as well as the different experiment data such as user custom tag, build Microblog user interest model, research and design the user interest model of multi-source data fusion, using space vector interest the final build.Different data sources is proposed in this paper Microblog users interested in fusion model, through the study found that can be effective use of interactive relationship between the user information for Microblog users interested in theme, and is superior to the effect of fusion model. The future through the user’s interest model can be targeted for Microblog users personalized recommendation, this model can be generalized to other social media sites platform, have certain business research value for enterprises.