Algorithms of BBS Opinion Leader Mining Based on Sentiment Analysis

A method to find the interest group based on topic content analysis, which combine the advantages of clustering and classification algorithms and the method of sentiment analysis to define the authority value as the weight of the link between users is proposed.

Abstract

Opinion leaders play a crucial role in online communities, which can guide the direction of public opinion. Most proposed algorithms on opinion leaders mining in internet social network are based on network structure and usually omit the fact that opinion leaders are field-limited and the opinion sentiment orientation analysis is the vital factor of one's authority. We propose a method to find the interest group based on topic content analysis, which combine the advantages of clustering and classification algorithms. Then we use the method of sentiment analysis to define the authority value as the weight of the link between users. On this basis, an algorithm named LeaderRank is proposed to identify the opinion leaders in BBS, and experiments indicate that LeaderRank algorithm can effectively improve the accuracy of leaders mining.