Knapsack-Based Reverse Influence Maximization for Target Marketing in Social Networks

With the dramatic proliferation in recent years, the social networks have become a ubiquitous medium of marketing and the influence maximization (IM) technique, being such a viral marketing tool, has gained significant research interest in recent years. The IM determines the influential users who maximize the profit defined by the maximum number of nodes that can be activated by a given seed set. However, most of the existing IM studies do not focus on estimating the seeding cost which is identified by the minimum number of nodes that must be activated in order to influence the given seed set. They either assume the seed nodes are initially activated, or some free products or services are offered to activate the seed nodes. However, seed users might also be activated by some other influential users, and thus, the reverse influence maximization (RIM) models have been proposed to find the seeding cost of target marketing. However, the existing RIM models are incapable of resolving the challenging issues and providing better seeding cost simultaneously. Therefore, in this paper, we propose a Knapsack-based solution (KRIM) under linear threshold (LT) model which not only resolves the RIM challenges efficiently, but also yields optimized seeding cost. The experimental results on both the synthesized and real datasets show that our model performs better than existing RIM models concerning estimated seeding cost, running time, and handling RIM-challenges.