Algorithm For Community Leader Detection In Social Network Services

The users link to each other by following and form communities insocial network. The detection of community leader is a topic of socialnetworks study. Leader detection is to identify the most important nodes inthe network, which involves the analysis of community centrality.Eigenvector centrality is one of the standards commonly used. In the socialnetwork, the connection between the users is by the following. In the Web,the connection between the pages is by the linking. There are similaritiesbetween them. This article makes improvements based on the PageRankalgorithm to generate a new algorithm UserRank for the detection ofcommunity leader. Traditional algorithm transfers the average of influenceto the every user followed. The new algorithm transfers the influence inproportion to the every user followed. In the community, a user’s influencetransfers to others: If user X follows to user Y, the influence of user Xtransfers to user Y fully or partly. After several algorithm iterations,influence of community user reaches to convergence and stable. The userwith the highest-ranking influence is the community leader. Experimentalresults show that the new algorithm can detect out the community leadersmore quickly and efficiently.