Detecting Opinion Leaders in Online Communities Based on an Improved PageRank Algorithm

Abstract

With the rapid development of web 2.0, online communities have become an important way to disclosure and spread the public sentiment. Opinion leaders in online communities play an important role during the formation of public opinion. The paper designs and implements an opinion leader detecting method based on an improved PageRank algorithm. The improved PageRank algorithm uses link relevance to define the weight of the link between users. The first step is crawling data from online communities and preprocessing them. Then construct the weight matrix by calculating link relevance between users. Finally, use the improved PageRank algorithm to rank users and detect opinion leaders. The experimental results show that, compared with the baselines, the proposed method can effectively identify opinion leaders in online communities.