Efficient Adoption Maximization in Multi-layer Social Networks

Influence maximization (IM) aims to find a set of influential users from social networks that can maximize the propagation of promoted information. Classical IM problem considers one social network only. While nowadays, people usually own several social network accounts. It is important to take advantage of multiple social networks when selecting influential users. In addition, even if a user is influenced by the information, because of personal preferences, there is still certain chance that he will change his mind about the information, i.e., not adopting the information. Therefore, to adapt to the real-world scenarios, we conduct the first research to investigate the adoption maximization problem in multi-layer social networks. We prove the problem is NP-hard. In order to handle large graphs efficiently, a h-hop based sampling method is developed, which can return a result with 1 - 1/e - ε approximation ratio. Experiments over real social networks are conducted to verify the advantages of proposed techniques.