Social Network Influence Maximization Based on Graph Attention Mechanisms

Key nodes in a social network have an important influence on the whole network, while some nodes in a social network may be at the center of the network while others may be at the edge of the network, traditional influence maximization algorithms ignore this imbalance, resulting in the dissemination of information by ignoring the nodes that are at the edge of the network but connected to the center. In order to avoid falling into a local optimal solution and failing to reach the global optimum so as to find the seed nodes with the highest influence, a graph attention mechanism based influence maximization algorithm for social networks (GATSH) is proposed. The GATSH algorithm is used on four public network datasets for experiments, and the results show that this method can effectively evaluate the importance of network nodes.