A Novel Approach for Influence Maximization Based on Clonal Selection Theory in Social Networks

More researchers have paid their attention to influence maximization which aims to find a k-size set of nodes to get maximum influence spread under a specific propagation model on social networks for recent years. However, the imbalance of accuracy and efficiency blocks the study of influence maximization for the last couple of years. For example, the greedy algorithm guarantee accuracy with expensive computation, while heuristic algorithms like degree heuristic algorithm are more efficiency but suffer from the unstable accuracy. In this paper, we study the clone selection algorithm based on clonal selection theory in immune system which motivates us to apply it to influence maximization. So, we propose a novel algorithm for influence maximization called CSAIM. We divide the original networks into communities with the method of community detection to reduce the scale of networks aiming to achieve remarkable efficiency and select superior nodes by eigenvector centrality to improve the quality of possible solutions which may improve accuracy. For possible solutions mentioned above, we simulate the process of immune system to generate the optimal solutions finally. The experimental results of the proposed algorithm on three public datasets demonstrate that the CSAIM can find the optimal solutions by fewer iterations which effectively promote the efficiency and keep the accuracy of influence spread.