SP-GCRL: Influence Maximization on Incomplete Social Graphs
Researchers have introduced SP-GCRL, a novel framework designed to tackle influence maximization challenges in social networks with incomplete data. The system employs a social-propagation-aware diffusion function to model complex user interactions and uses contrastive learning to generate robust node representations, even with missing connections. By integrating a graph attention network for efficient surrogate metric calculation and employing DDQN for policy learning, SP-GCRL demonstrates significant improvements over existing methods on real-world networks. AI