Dynamic Graph Transformer with Correlated Spatial-Temporal Positional Encoding (CorDGT)

This is the official implementation of the WSDM 2025 paper:

Zhe Wang, Sheng Zhou, Jiawei Chen, Zhen Zhang, Binbin Hu, Yan Feng, Chun Chen, Can Wang. Dynamic Graph Transformer with Correlated Spatial-Temporal Positional Encoding. [arXiv link]

Datasets Downloading and Preprocessing

All the datasets can be downloaded here. Put datasets into processed folder, then run the following command to preprocess the datasets:

python process.py 

We use the dense npy format to save the features in binary format. If edge features or nodes features are absent, it will be replaced by a vector of zeros.

Training and Evaluation Commands

• Use CorDGT for Dynamic Link Prediciton:

# Enron 
python -u learn_edge.py -d enron --uniform --bs 100 --n_degree 20 1 --n_head 6

# UCI
python -u learn_edge.py -d uci --bs 100 --uniform --n_degree 32 1 --n_head 6 --alpha 1 --beta 0.1 

Requirements

• python >= 3.8.0
• torch >= 1.9.1
• Full Dependency list is in requirements.txt

Acknowledgments

Some of this code are based on TGAT. We are grateful for authors' contributions.

Citations

If you find this work useful in your research, please consider citing:

@inproceedings{wang2024dynamic,
  title={Dynamic Graph Transformer with Correlated Spatial-Temporal Positional Encoding},
  author={Wang, Zhe and Zhou, Sheng and Chen, Jiawei and Zhang, Zhen and Hu, Binbin and Feng, Yan and Chen, Chun and Wang, Can},
  booktitle={Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining},
  year={2025}
}