Reverse Path Activation-based Reverse Influence Maximization in Social Networks

Ashis Talukder, Anupam Kumar Bairagi, Do Hyeon Kim, Choong Seon Hong

http://doi.org/10.5626/JOK.2018.45.11.1203

Influence Maximization (IM) deals with finding influential users for viral marketing in social networks, whereas Reverse Influence Maximization (RIM), a new research direction in the influence-maximization domain, deals with seeding cost, also known as opportunity cost. The IM estimates a small seed set in such a way that by targeting those seed nodes, the influence is maximized in the network. Generally, the seed nodes are assumed to be activated initially in the IM problem. However, we argue that seed nodes need to be influenced by some of their in-neighbor nodes in a similar way how an activated node influences its out-neighbors to be activated. The RIM problem finds the seeding cost, which is defined by the minimum number of nodes that must be activated in order to activate all the seed nodes. In this paper, we propose an Active Reverse Path-based Reverse Influence Maximization (ARP-RIM) model to find the minimum seeding cost. Our model is based on the voting model and the classic Independent Cascade model. We simulate our model with three real datasets of three popular social networks. The experimental result shows that the ARP-RIM model outperforms existing RIM models.