Influence maximization based on double clusters multi-verse optimizer

Abstract

The influence maximization problem is one of the important research topics in the social network. We address the problems of meta-heuristic algorithms such as the high probability of entrapment in local optima, low accuracy of the solution, and decreasing diversity in the late iteration. The Double Clusters Multi-Verse Optimizer (DCMVO) algorithm is proposed as a solution to the problem of influence maximization. In DCMVO, based on the fact that nodes in the social network are susceptible to the influence of neighboring nodes, individuals in the globular cluster are updated using neighboring nodes with high similarity, which enhances local exploitation and improves the accuracy of the solution. To improve the global exploration, using a comprehensive learning strategy in the open cluster enables individuals to learn from surrounding individuals by dimensions, thereby expanding the search space. The wormhole mechanism is used to enhance the information interaction between double clusters during the iterative process, which serves to balance local exploitation and global exploration. Under the independent cascade (IC) model, extensive experiments conducted on seven actual social networks demonstrate the effectiveness of DCMVO.