Enhancing Heterogeneous Graph Representation Learning through Generative Self-Supervised Approach

A generative self-supervised learning approach, HGVAE, is proposed to effectively refine the latent representations of heterogeneous graphs, leading to improved performance on downstream tasks.