A moth-flame algorithm based on social learning theory for influence maximization

Swarm intelligent optimization algorithm plays a key role in solving the influence maximization problem. Aiming at the problems of insufficient diversity and overlapping influence when selecting initial nodes in previous influence maximization algorithms, the local influence measure of nodes is constructed considering the network topology, and the population initialization method based on local influence is proposed. Aiming at the traditional moth-flame algorithm which is difficult to deal with the balance between global exploration and local exploitation, resulting in the decline of population diversity and easy to fall into the local optimal solution, combining with the social learning theory, re-planning the population state, defining multiple learning strategies for individuals in different states, and proposing a new method for the moth-flame algorithm based on the social learning theory to solve the problem of maximizing the influence of the social network. Experiments are conducted on six real social networks under the independent cascade model (IC), and the results show that the proposed algorithm can effectively identify the influential node sets.