Effective Approaches for Influence Maximization in Social Networks

Abstract

The detection of the top influential users is well-known scientifically as the social influence maximization. The current existing solutions suffer from several limitations, such as the highly required computations and the running time to find the top influential seed set. Therefore, finding an effective and efficient solution is still a challenging task. In order to solve the current scientific gap, this thesis proposes an effective and scalable community-based approach for the influence maximization problem called Louvain-k-shell Generalization (LKG). LKG is a fast and scalable community-based hybrid approach to detect top influential users in social networks. The LKG hybrid approach consists of three phases: 1) Community detection, in which the complete social network is partitioned into related communities using the Louvain algorithm; 2) Community top nodes detection that applies the k-shell decomposition locally in each portioned community; and finally 3) Selection generalization, in which the prior obtained results are generalized over the whole network for maximizing the global spread of influence. The results of the LKG approach have been shown to achieve better results for the spread of influence using incomplete social networks than the existing related work approaches and with far much less processing time.