Identifying and evaluating the internet opinion leader community based on k-clique clustering

With the rapid development of the Internet, the Internet has become an important place for producing and spreading public sentiment. Opinion leaders play an important role in leading the public opinion. In this paper, we extract the communities by analyzing the replies of each post in the bulletin board system. Then, an opinion leader community mining method is proposed based on the level structure. Thus, the communities have better overlaps and multiple relations. Also, we analyze the revolution of the opinion leader communities and put forward a time-dividing method. With this method, we divided whole communities into different pieces based on the character of the post and the duration of the time. And we come up with suitable parameters to receive the evolution results of the communities. Finally, our experiments prove the efficiency of the opinion leader community mining method, and we summarize the properties of the opinion leader community in revolution.