基于重力模型和图注意力网络的重要节点排序方法OA

Ranking important nodes in complex networks enables the identification of core and critical nodes,which plays an important role in controlling the propagation process that occurs on complex networks.There have been many studies on the ranking of important nodes,but further exploration is needed on how to efficiently and comprehensively utilize node information to extract features and evaluate node importance.A method for ranking important nodes based on graph neural network(GNN)was proposed by combining local and global information of nodes.This method first captured local and global information of nodes using degree,shortest distance between nodes,and k-shell value.Subsequently,after obtaining node information,the method used graph attention mechanism to aggregate neighboring node features and learn node representation vectors.Finally,the importance of nodes was calculated through a fully connected layer.Experiments were conducted on real network datasets and the effectiveness of the proposed method was validated by using monotonicity index and Kendall τ coefficient.The experimental results show that the model can effectively sort nodes in different network structures.