Minimizing the seed set cost for influence spreading with the probabilistic guarantee

Following the fast growth of the Internet, the issue of influence maximization (IM) in social network has aroused great interest of the researchers and has been applied in many areas such as marketing and finance. In real world applications, the promulgator wants that the propagating cost is minimized while the percentage of the influenced users in a given range reaches a desired threshold under a predefined probability. This paper first defines this problem as minimum cost seed selection with probabilistic influence spreading guarantee (MCSS-PISG) in linear threshold (LT) model. We prove that the problem MCSS-PISG is NP-hard and its influence function is monotonous and sub-modular. To avoid simulating the influence propagation, which requires large amount of computation time, an algorithm is presented for estimating the influence propagation by path counting in the sampled graphs. An algorithm named Sampling_Greedy is also proposed for MCSS-PISG problem, and error of the result by Sampling_Greedy is analyzed. Experimental results on real world networks indicate that algorithm Sampling_Greedy has higher performance than other methods. (C) 2021 Elsevier B.V. All rights reserved.