Research On Microblog User's Interest Mining Based On User Focus_Author Topic Model

MicroBlog is cherished by people in the web2.0 era,as a community that gathering interest attracting billions of users.Due to its ease of operation, releasing MicroBlog frequently, and spread fast with wide range of.In a mass amount of MicroBlog information, Mining accurate user interest and providing satisfactory ser-vice to users has important meaning.Traditional clustering methods, vector space approach are not suitable for processing with MicroBlog data with the characteristics that are noisy and is not standardized, wide dimension. Though the topic model approach which applied to the user theme mining can obtain some user interest, it has not combined the Microblog data feature and theusing feature.To solve the above problems, integrate the original and forwarding Microblog, using the author topic model mining the topic of user interest, combined the MicroBlog user'focused feature improved the author topic model providing the user interest mining method which is called UF_AT model, through progressive layers method obtaining the accurate user interesting. Through experiment capture data and mining user'interest, Compared the experimental result on the basic of the probability of user with topics and topic with words,through the evaluation index of perplexity and precision prove the User focus Author topic model has better effect in mining the MicroBlog user'interest.