Nodes Grouping Genetic Algorithm for Influence Maximization in Multiplex Social Networks

Abstract

Influence maximization (IM) aims to select a small number of seed users who can maximize the influence of information spread in social networks. The influence maximization problem in multiplex social networks considers the effects of overlapping users between different social networks on spreading the influence across networks. Since nodes in the network have different selection cost, the importance of a node cannot be determined only by the node's influence. This paper proposes a genetic algorithm using a novel node grouping strategy based on the node influence and selection cost, termed NGGA, for multiplex social networks. A node selection operation uses a shielding node set to realize a flexible search. Experimental results on three real multiplex networks demonstrate the effectiveness of the proposed algorithm.