TREND: TempoRal Event and Node Dynamics for Graph Representation Learning

We provide the implementaion of TREND model, which is the source code for the WWW 2022 paper "TREND: TempoRal Event and Node Dynamics for Graph Representation Learning".

The repository is organised as follows:

• dataset/: the directory of data sets, and it contains the cit-HepTh data set as the example. You can download the other two datasets wiki and Taobao, through the google drive link: https://drive.google.com/drive/mobile/folders/19tcuesVuPpVM0vV96DuytngPQ_JMt_fe?pli=1&sort=13&direction=a
• res/: the directory of saved models.
• Emlp.py: the transfer function for Hawkes process.
• data_dyn_cite.py: training data preprocessing.
• data_tlp_cite.py: testing data preperation.
• dgnn.py: the Hawkes process based GNN.
• film.py: the event-conditioned transformation.
• main_test: the testing entrance.
• main_train: the training entrance.
• model: the whole model of proposed TREND.
• node_relu: the MLP of node-dynamics predictor.

Requirements

To install requirements:

pip install -r requirements.txt

Train and test

To train the model in the paper:

python main_train.py

To test the trained model:

python main_test.py

Cite

@inproceedings{wen2022trend,
	title = {TREND: TempoRal Event and Node Dynamics for Graph Representation Learning},
	author = {Wen, Zhihao and Fang, Yuan},
	booktitle = {Proceedings of the Web Conference 2022},
	year = {2022}
}