Heterogeneous Graph Edge Prediction using RelGraphConv

Hi all,

I was wondering if there has been a DGL implementation of Heterogeneous Graph Edge Prediction for graphs with multiple edge and node types, preferably using RelGraphConv layers, that has been made public? I would like to see an example of such a problem before I begin attempting to build such a model for a complex knowledge base recommendation system.

Additionally, if there isn’t such implementation available, would there be a small community of developers whom are currently researching this specific task for heterogenous graphs? I’m interested in discussing the task further and would enjoy knowing whom else is interested in this topic.

OpenHGNN (GitHub - BUPT-GAMMA/OpenHGNN: This is an open-source toolkit for Heterogeneous Graph Neural Network(OpenHGNN) based on DGL.) has a bunch of heterogeneous graph link prediction examples.

1 Like