Research On Credibility Of Microblog Users Based On Link Analysis

As a new platform, microblog becomes more and more popular. Nowadays, it has been an important part of people’s lives to share and receive information by microblog. However, due to its spontaneity, low limit and lack of effective management means to users, the microblog is full of false information and rumors along with the increase in the number of users. Therefore, the user’s credibility gradually arouses people’s attention and becomes a social problem need to be solved.Messages on microblog belong to a part of the network information, the main challenges of research on microblog credibility are the small size of messages and the complex features of microblog, and how to accurately assess the credibility of users has become a research hotspot. At present, the credibility research has achieved some initial results, but with the continuous development of network technology, social practices have also been gradually changed, the traditional credibility evaluation methods are insufficient to meet the needs of new social network and have appeared many shortcomings, such as single criteria, time-consuming process and characteristics considered incompletely. Based on the above background, with SINA microblog as the research object, the thesis starts from the basic theory of credibility and social network to analyze the trust relationship between users. And then, on the basis of PageRank idea, combined with user’s attributes, the thesis puts forward relevant model and algorithm to measure and rank the credibility of the user. The main content of this thesis is as follows:First of all, the thesis illustrates the structure and characteristics of microblog, and analyzes the impact of new communication mode to user’s behavior. Next, it statements related theory and the key technology about the credibility, and then discusses the research content of microblog credibility. After that, the thesis analyzes existing research models and methods about microblog credibility which can provide some useful reference for this study.Secondly, based on the dissemination mechanism of message on microblog and equilibrium theory of social network, the thesis analyzes the transitive trust relationship between users and points out that similar users on credibility are more likely to establish connection in stable microblog network. Adopting the principle of PageRank, the thesis transforms the user’s credibility problem into the problem to calculate or evaluate support from other users with the consideration of links among users. On this basis, it builds model about credibility evaluation from the perspective of link analysis.Finally, the thesis analyzes credibility ranking algorithm, which takes advantage of the idea of uneven weights and particularly emphasizes the importance of fans’ quality, can effectively reduce the impact of “zombie fans” in the process of sorting. The experimental results over real data set illustrate the effectiveness of the proposed algorithm.