PyToch implementation of R-GCN model for node classification and link prediction
- Code skeleton
- Get link_data to DistMul properly
- Get the adjacency matrix to DistMul
- Pipeline the two experiments somehow
- Test training loop for classification task
- Test training loop for link prediction
- Complete the evaluation scripts to save checkpoints and best model
- Evaluation scripts
- Revamp GCN into configurable layers
- Refactor into Encoder and Decoder hierarchical structure with Aggregator
Requirements: rdflib, wget, scipy, sklearn