Authors: Meng Liu, Yue Liu, Ke Liang, Wenxuan Tu, Siwei Wang, Sihang Zhou, Xinwang Liu
Paper: https://arxiv.org/abs/2305.10738 (ICLR 2024)
This is the PyTorch version of TGC. We want to provide you with as much usable code as possible.
If you find any problems, feel free to contact us: mengliuedu@163.com.
To run the code, you need prepare datasets and pretrain embeddings:
You can download the datasets from Data4TGC and create "data" folder in the same directory as the "emb" and "framework" folders.
In ./framework/pretrain/, you need run the pretrain.py to generate pretrain embeddings.
Note that these embeddings are used for TGC training, while the features in the dataset are used for training by any other method.
That is, the pre-training of node2vec is only part of the TGC.
You need create a folder for each dataset in ./emb/ to store generated node embeddings.
For example, after training with Patent dataset, the node embeddings will be stored in ./emb/patent/
For each dataset, create a folder in emb folder with its corresponding name to store node embeddings, i.e., for arXivAI dataset, create ./emb/arXivAI.
For training, run the main.py in the ./framework folder, all parameter settings have default values, you can adjust them in main.py.
For test, you have two ways:
(1) In the training process, we evaluate the clustering performance for each epoch.
(2) You can also run the clustering.py in the ./framework/experiments folder.
Note that the node embeddings in the ./emb./patent/patent_TGC_200.emb folder are just placeholders, you need to run the main code to generate them.
If you feel our work has been helpful, thank you for the citation.
@inproceedings{TGC_ML_ICLR,
title={Deep Temporal Graph Clustering},
author={Liu, Meng and Liu, Yue and Liang, Ke and Tu, Wenxuan and Wang, Siwei and Zhou, Sihang and Liu, Xinwang},
booktitle={The 12th International Conference on Learning Representations},
year={2024}
}