Community Centrality-Based Greedy Approach for Identifying Top- Influencers in Social Networks

Online social network today is an effective media to share and disperse tons of information, especially for advertizing and marketing. However, with limited budgets, commercial companies make hard efforts to determine a set of source persons who can highly diffuse information of their products, implying that more benefits will be received. In this paper, we propose an algorithm, called community centrality-based greedy algorithm, for the problem of finding top- influencers in social networks. The algorithm is composed of four main processes. First, a social network is partitioned into communities using the Markov clustering algorithm. Second, nodes with highest centrality values are extracted from each community. Third, some communities are combined; and last, top- influencers are determined from a set of highest centrality nodes based on the independent cascade model. We conduct experiments on a publicly available Higgs Twitter dataset. Experimental results show that the proposed algorithm executes much faster than the state-of-the-art greedy one, while still maximized nearly the same influence spread.