Maximal Clique Based Influence Maximization in Networks

Influence maximization is a fundamental problem in several real life applications such as viral marketing, recommendation system, collaboration and social networks. Maximizing influence spreading in a given network aims to find the initially active vertex set of size k called seed nodes (or initial spreaders (In this paper, we use seed set and initial spreaders interchangeably.)) which maximizes the expected number of the infected vertices. The state-of-the-art local-based techniques developed to solve this problem are based on local structure information such as degree centrality, nodes clustering coefficient, and others utilize the whole network structure, such as k-core decomposition, and node betweenness. In this paper, we aim at solving the problem of influence maximization using maximal clique problem. Our intuition is based on the fact that the presence of a dense neighborhood around a node is fundamental to the maximization of influence. Our approach follows the following three steps: (1) discovering all the maximal cliques from the complex network; (2) filtering the set of maximal cliques; we then denote the vertices belonging to the rest of maximal cliques as superordinate vertices, and (3) ranking the superordinate nodes according to some indicators. We evaluate the proposed framework empirically against several high-performing methods on a number of real-life datasets. The experimental results show that our algorithms outperform existing state-of-the-art methods in finding the best initial spreaders in networks.