Research on Social Network Influence Maximization Algorithm Based on Graph Neural Network

In this paper, we propose an innovative Graph Neural Network (GNN) model that combines PageRank, genetic algorithm, and Graph Convolutional neural Network (GCN) to solve the problem of influence maximization in social networks. By utilizing the PageRank algorithm to pre-rank the nodes, we obtain the potential key influencers in the network. Subsequently, genetic algorithm is introduced to optimize the selection of seed nodes, and the optimal combination of nodes is found through an iterative evolution process. On this basis, we apply GCN to capture and analyze the complex node relationships and attribute information in social networks to more accurately predict the efficiency and scope of information dissemination. GCN not only considers the direct connection of nodes, but also realizes the modeling of long-distance dependencies through multi-layer information transfer mechanism. Finally, through the joint use of the node set optimized by genetic algorithm and GCN, the performance of the social network influence maximization task is significantly improved, which provides a strong theoretical support and technical solution for precision marketing and social recommendation system. This study extends the application boundaries of GNNS in the analysis of real-world complex networks and demonstrates their potential in processing large-scale, high-dimensional datasets.