Hy-Transformer

This repository is the official implementation of Improving Hyper-Relational Knowledge Graph Completion in AAAI 2022 Deep Learning on Graphs Workshop. The majority of this code base is from https://github.com/migalkin/StarE. We thank the authors for their contribution.

Requirements

• Python 3.7
• PyTorch 1.5.1
• torch-geometric 1.6.1
• torch-scatter 2.0.5
• tqdm
• wandb

Create a new conda environment and execute setup.sh. Alternatively

pip install -r requirements.txt

Running the experiments

Running on WD50K dataset:

./run_batch_wd50k.sh ${GPU-ID} wd50k statements StarE hy-transformer_mask True

Running on JF17K dataset:

./run_batch_jf17k.sh ${GPU-ID} jf17k statements StarE hy-transformer_mask True 

Running on WikiPeople dataset:

./run_batch_wikipeople.sh ${GPU-ID} wikipeople statements StarE hy-transformer_mask True

For any further questions, please contact: [email protected]