mertkosan / DyGED

Dynamic Graph Event Detection
MIT License
5 stars 3 forks source link

Event Detection on Dynamic Graphs

This is a PyTorch implementation of DyGED (Dynamic Graph Event Detection) that is proposed event detection architecture in our paper:

Event Detection on Dynamic Graphs.
Mert Kosan, Arlei Silva, Sourav Medya, Brian Uzzi, Ambuj Singh.
Deep Learning on Graphs: Method and Applications, Association for the Advancement of Artificial Intelligence 2023 (DLG-AAAI’23).
Workshop version: https://drive.google.com/file/d/1ijb9ngmi-yaNAmtRi_9Bc7B9INn8Q_Ir/view
Longer version: https://arxiv.org/abs/2110.12148

DLG-AAAI'23 Poster

PDF version

Data

We share Twitter Weather and NYC Cab datasets in DyGED_data.zip. Hedge Fund data unfortunately cannot be publicized. We also provide data_utils.py which process the graphs into PyTorch Geometric format. Please contact us if any issue encountered.

Implementation

We share our framework in models.py, but it does not support PyTorch geometric, but raw data. We hope to implement PyTorch Geometric version soon.

Citing

If you find our framework or data useful, please consider citing the following paper:

@article{kosan2021event,
  title={Event detection on dynamic graphs},
  author={Kosan, Mert and Silva, Arlei and Medya, Sourav and Uzzi, Brian and Singh, Ambuj},
  journal={arXiv preprint arXiv:2110.12148},
  year={2021}
}