Research And Application Of Micro-blog User Interest Model And Personalized Recommendation Technique

Along with the speedy development of the Internet, the network plays an increasing important role in our daily life. As time passed, the explosion of Internet information and internet user group scale processed makes an increasingly diverse on users’needs. It causes personalization recommendation system to become a research hotpot. In order to recommend the information to Internet users quickly and effectively, getting the users’ accurate interests becomes the most important aspect. With arriving of the Web2.0 time, the new social media with a stronger interaction, such as Micro-blog, is becoming increasingly popular. Micro-blog includes such traits as low threshold, simple use, which make it could gather a large amount of users in a short period. The mass data of micro-blog provides us a new data pool to analyze users’interests for personalization recommendation. The thesis is focus on the research of micro-blog and studies the technology for user interest mining and personalized recommendation.In the first part, this thesis studies the information from Micro-blog and user interest model. The thesis is based on the information feature of Micro-blog, and the research included three main aspects: the extracting mode of users’interests, the representation method of interest model and update methods. Then it works out the user interest model of Micro-blog which is based on the LDA topic model. Meanwhile, the experiment shows that the model perfectly agrees with the manual output results, and it also has a good accuracy.Secondly, on the basis of micro-blog user interest model, the author provides the study of the personalized recommendation technique. The first step is to give out an overview of several mainstream technologies and analysis advantages and disadvantages with the scope of them. Then, the author offeres a personalization recommendation algorithm which is based on the collaborative filtering and LDA topic model to verify the feasibility of the method mentioned in this thesis.The finally part of the thesis applies the theoretical study results to a real personalized recommender system. The whole system framework design and the implementation details about how to achieve key modules of the system would also include in this part.