Research on methods to identify the opinion leaders in internet community

Abstract

With the rapid development of WEB2.0, Internet forums have become an information exchanged platform where netizens can freely express their opinions. In Internet consensus environment, it is very important to appropriately control and lead the consensus and the opinion leaders play critical roles in Internet community. This paper, by combining the characteristics of Internet forums, established a PageRank Model to identify the opinion leaders based on the influence of linguistic contents and the sentiment tendentiousness. The PageRank Model adopts the product of the influences of the contents and sentiments (i.e., integrated influence) as the weighted value of two ID numbers and achieves the PR value via the PageRank calculation, thereby identifying the opinion leader of the forum. Tests have verified that the original PageRank vector of the Inlink has its advantages, compared with other methods, the ID of the opinion leader's articles identified by the PageRank Model can cover 80% of the forum IDs, which is in compliance with the 2-8 law. Therefore the algorithms based on the PageRank Model can identify effectively the opinion leaders in Internet community.