Identification of Opinion Leaders in Social Networks Based on Sentiment Analysis: Evidence from an Automotive Forum

摘要

Opinion leadership of social network plays an important role in the fields of knowledge spreading, public sentiment controlling, marketing, etc. The opinions of users are derived from the reviews of topics, and the analysis of users' sentiment is helpful in recognizing users' emotional preference of opinion leaders. Thus, it is necessary to classify the forum posts and refine the posts with highly professional knowledge. We then improve the sentiment analysis due to the imbalanced datasets, and establish a comprehensive attention and emotion weight matrix. Accordingly, in this paper, we are going to propose a Leader-PageRank algorithm, which is based on the social network structure and emotional tendency. We do the comparative experiments on the automotive forum, and the results show that the Leader-PageRank algorithm can identify the positive opinion leaders in the professional fields effectively through connecting with the interactions in social networks.