A Hybrid Community-based Simulated Annealing Approach for Influence Maximization in Social Networks

Abstract

Influence maximization (IM) in social networks aims to figure out the best subset of seed nodes which have maximum cascading influence under a diffusion model. This paper proposes a hybrid Community-based Simulated Annealing (ComSA) approach for the IM problem. A community detection algorithm is employed to segregate the entire social network structure into some more deeply clustered communities. Thereafter, a degree-based metric has been used to select the candidate pool from each community by excluding less influential nodes at the preliminary data preprocessing phase. A community-based seed initialization and neighborhood search technique have been proposed. To speed up the convergence of stable solutions in Simulated Annealing approach, a greedy hill climbing strategy is also implemented instead of using probabilistic based solution acceptance processes. Experimental results on four real-world datasets show that our proposed algorithm has comparable solution with greedy and outperforms the other existing meta-heuristic approaches.