Study Of Recommendation System Based On User Relationship Analysis And Micro-Blog Content Mining

With the coming era of web2.0, social network is developing rapidly. It gathers various kinds of people together with the help of internet, the new style of carrier, which has a significant impact on the information obtaining methods and life styles of individuals. In the recent two years, micro-blog, as a novel social network platform, appears in the sights of people, and it is widely popular in various kinds of users because of its novel and convenient model of information propagation. The rapid growth of the user number also leads to the explosive growth of the amount of the information. Therefore, how to extract the topics from massive amounts of information, which the user is interested in, and recommend the corresponding topics to the user is a research problem demanding to be solved immediately. This thesis is based on analysis of user relationship and mining of micro-blog contents to recommend information, and the purpose is to provide personalized service. The main work of this thesis includes:1. With analysis of user relationship, this thesis proposes an algorithm to detect influence value of related users to the core user. With the SINA micro-blog user data, this thesis analyzes the reason of formation of user relationship network. Combining with the fuzzy comprehensive evaluation method, we can confirm the relationships between factors which impact the influence value of users. And we propose a formula for calculating the influence value of user to build the model of user influence value for the core user..2. With analysis of micro-blog contents, this thesis proposes the research method standing on the level of topics of micro-blogs. Analysis of micro-blog contents. The thesis combines the credible association rules to detect the topics of related users. At the same time, we use theory of word activation forces to analyze the emotional tendentiousness of related user to identify his or her sentiment about a specific topic.3. Combining with the previous sections, this thesis provides a recommendation algorithm for micro-blog information. This thesis reviews some existing theories about information recommendation, and summarizes the recommendation methods of micro-blog information. Moreover, we propose an algorithm based on the analysis of user relationship and micro-blog contents, which can help to recommend micro-blogs to the core users and we prove it with experiment. This thesis is based on user relationship analysis and micro-blogscontents mining to recommend information, which can help to providepersonalized service. Good user experience could help to increase thesatisfaction of the user, and it can provide theory basis for public opinionsmonitoring, which has both theoretical value and practical value.