Research Of Opinion Leaders Mining Method In Social Network

Social network is an online platform without boundaries. Users who own the same knowledge structure and interests in same field can take part in this platform which is helpful for information dessemination. In the process of information dissemination, opinion leaders are the ones who play an increasingly important influence. Therefore, the influence of opinion leaders is widely used in many fields, such as public opinion monitoring, information promotion, e-commerce and other fields. How to find a user set with some influence accurately in social network with large scale data has always been the research object of opinion leaders mining. At present, there are three main kinds of opinion leaders mining methods, namely, mining method based on statistics, network structure analysis and cluster analysis. Among these methods, PageRank algorithm based on network structure can quickly handle large amounts of data. The classic algorithm of web page ranking is also used to calculate the influence of nodes. But in the PageRank algorithm, the value of the page's influence is transmitted to the chain out of the page, without taking into account the differences between the pages of the vote, that is, the user's behavior differences.However, the behaviors of network users often reflect their true thoughts and ideas. Users will make friends with the other users who are the old friends from their real life or unfamiliar users in the social network. After becoming friends, users can pay attention to all kinds of information released by their friends. They choose to poste comments or forwarding behavior according to their preferences. Therefore, the definition of attention behavior and attention is drawn in this thesis on the basis of throughing the analysis of user behavior in social networks, and discusses the influence of user's behavior on the user's influence. Secondly, considering two aspects of the network structure and user behavior, an opinion leaders mining algorithm called SNURank in social network is proposed in this thesis, including the idea of PageRank algorithm. It defines the formula of attention degree based on the analysis of user attention behavior, and this can be used to calculate the influence of users to find opinion leaders finally.At last, the experimental analysis of the data set from sina micro-blog shows that the algorithm is effective and can identify opinion leaders accurately.