A Two-Stage Seeds Algorithm for Competitive Influence Maximization Considering User Demand

Competitive influence maximization (CIM) in online social network has received widespread attention and research in recent years. The traditional competitive influence maximization problem explores the competition between multiple entities on the same network, aiming to select a certain number of seeds for one of the entities to maximize the spread of its influence. However, the latest competitive influence maximization researches ignore the impact of differences in user demand on the spread process of competitive influence in real-world competitive relationships. Therefore, a novel propagation model called Competitive Linear Threshold Model considering User Demand (CLTMcUD) is presented, which takes into account the difference in user demand for two different brands of the same product category. For this propagation model, a two-stage algorithm named Dual Influence Assessment based on Community Structure (DIACS) algorithm is proposed, which utilizes the characteristics of community structure and dual-influence of nodes to select candidate seeds and maximize the influence of a competitor. We test our algorithm on four real-world datasets and show that it outperforms state-of-the-art algorithms.