Abstract:

We use a method of social network analysis to indentify opinion leaders in social networks. First of all we analyze the characteristics of the social networks of online communities, consider the difference between BBS, blogosphere and Q&A network, and propose that undirected weighted network models will enhance the identification accuracy of opinion leaders. Then, we analyze the features of social networks. We validate that BBS is smallworld and scalefree, and we make a quantitative analysis of the social background of opinion leaders. We also propose a PageRank algorithm based on undirected weighted networks, and test several algorithms proposed before our experiment. We use four years’ data from a BBS, and test these algorithms. Finally, we make  a profound analysis and find the relationship between opinion leaders and boards, and the result shows that we can find a majority of opinion leaders from covering few boards.

Key words: complex networks;online community;opinion leader;PageRank