This project implements the code for the paper "Unsupervised Graph Clustering with Deep Structural Entropy", which is accepted to Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining 2025 (KDD 2025).

If you use this project, please cite:

@inproceedings{zhang2025unsupervised, 
    title      ={Unsupervised Graph Clustering with Deep Structural Entropy}, 
    author     ={Zhang, Jingyun and Peng, Hao and Sun, Li and Wu, Guanlin and Liu, Chunyang and Yu, Zhengtao}, 
    booktitle  ={Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2 (KDD ’25)},
    year       ={2025}, 
    publisher  ={ACM},
    doi        ={10.1145/3711896.3737173},
    url        ={https://doi.org/10.1145/3711896.3737173}
    } 

If you have any questions, feel free to open an issue or email to: BY2339214@buaa.edu.cn

Environment & Requirements

 -python 3.10.12
 -torch 2.3.1+cu118
 -torch_scatter 2.1.2+pt23cu118
 -torch_geometric 2.5.3
 -numpy 1.26.3
 -dgl 2.3.0+cu118
 -scipy 1.14.0
 -munkres 1.1.4

Project Structure

main.py: Entry point for training or evaluation.

model.py: GCN, ASS, and DeSE models.

utility/: Utility functions.

└── dataset.py: Data loading and preprocessing.

└── parser.py: Parameters setting and description.

└── util.py: cluster metrics and conversion functions.

Quick Start

python DeSE/main.py