Internet Opinion Leader Detection Based On Text Sentiment Analysis

Opinion leader means a person who can offer others information, opinion, comment and effect other people via social network. In Internet public opinion analysis, opinion leader can spread information, enlarge influence and guide public opinion. Nowadays, method of identifying Internet opinion leader concentrates on social network analysis and clustering analysis. These methods only focus on relationships between users without users’own information and comments information. Besides, some bloggers gain large amount of reviews which are full of criticizes. In this case, these bloggers don’t have positive effects of opinion leader as public opinion monitor or guidance. Text sentiment analysis can detect that a comment is positive, negative or neutrality. So we can remove some famous bloggers who receive a lot of negative comments. Index analysis can integrate all the necessary indexes including users’personal information and relationships among bloggers. So it gives a dependable evidence to find the real opinion leader.The paper based on text sentiment analysis and index analysis, which includes constructing indexes, calculating weigh and ranking. Then the paper uses Word2Vec algorithm to find new Internet sentiment words to improve the accuracy rate of text sentiment analysis which can help us to discriminate text into positive attitude, negative attitude or neutrality attitude. So the contentious bloggers can be detected and removed. Thus we compare opinion leaders in our method with ones in WeiboRank (based on PageRank) method to prove our method’s effectiveness and reliability. The main contribution of this paper is identifying the three kinds of typical opinion leaders, including the whole process of public opinion spread modules from the origin of emergencies, expending influence to the public, and attitude changing. In this case, our work can monitor network public opinion evolution, capture and forecast people’s viewpoint, and can be used as an important measure to prevent group radical events.