Influence Maximization Based on True Threshold in Social Networks

As E-marketing based on online social networks develops fast, the influence maximization problem draws attention from both academics and industries. This problem focuses on which subset of users should be selected as seed users so that based on the specific information diffusion model, the advertising companies can maximize word-of-mouth effect. Exisiting related work assume there is no cost to choose these seed users, or the cost is given in the problem setting. While in real situation, it is crucial but difficult to elicit users’ true attitude over being seeds. Moreover, we notice “threshold” as users’ private information in the Linear Threshold model can represent individual’s preference. Thus we propose a new model in which users, willing to be seeds, are asked to report their threshold information. The method called TREE is designed to solve this model, especially the payment mechanism should make sure all users tell truth. Experiments on real social network data to verify the effectiveness of TREE.