Detecting Influential Users in Social Networks Based on Bipartite Comments Graph

Abstract

With the development of online social networks, the task of identifying users who have a great influence on other participants in social networks is becoming increasingly important. An important source of information is user comments on content created by other users. The paper proposes a method for determining influence based on a bipartite user-comment-content graph. It incorporates information about text messages and the reactions of other users to them. In addition, we propose a method for identifying user communities in such a graph based on common interests. Experiments on data collections from VKontakte and YouTube networks show the correlation between user activity and influence, however, the most active commenters are not necessarily the most influential. Community analysis shows a positive correlation between the size of a community, the number of most influential users in it, and the average influence of community users.