gIM: GPU Accelerated RIS-Based Influence Maximization Algorithm

Given a social network modeled as weighted graph GG, the influence maximization problem seeks kk vertices to become initially influenced, maximize expected number of influenced nodes under particular diffusion model. The has been proven be NP-hard, and most proposed solutions are approximate greedy algorithms, which can guarantee tunable approximation ratio for their results with respect optimal solution. state-of-the-art algorithms based on Reverse Influence Sampling (RIS) technique, offer both computational efficiency non-trivial (1-1/e-?)(1-1/e-?)-approximation any ? > 0?>0. RIS-based despite lower cost compared other methods, still require long running times solve in large-scale graphs low values ?. In this article, we present novel efficient parallel implementation algorithm, namely IMM, GPU. GPU-accelerated named gIM, significantly reduce time Furthermore, show that gIM algorithm variations IM problem, only by applying minor modifications. Experimental solution reduces runtime factor up 220 ×. source code is publicly available online.