NLA-GNN: Non-local information aggregated graph neural network for heterogeneous graph embedding

Siheng Wang, Guitao Cao, Wenming Cao, Yan Li

Abstract: Highlights•NLA-GNN is a new framework to generate node embeddings from heterogeneous graphs.•NLA-GNN simultaneously considers local information and non-local information.•We propose a two-step method to enable non-local aggregation for heterogeneous graphs.