Embedding Node Structural Role Identity Using Stress Majorization

Requirements:

env.yml

How to use:

Our model stress\models\sm2vec.py

vis_barbell.py embedding visualization on barbell graph

vis_karate.py embedding visualization on karate network

evaluate_synthetic.py node clustering experiments on synthetic graphs (default: 'varied' setting)

evaluate_brazil.py node classification experiments on the brazil air-traffic networks

evaluate_europe.py node classification experiments on the europe air-traffic networks

evaluate_usa.py node classification experiments on the usa air-traffic networks

Code reference:

This code is built upon the code of the following paper:

Claire Donnat, Marinka Zitnik, David Hallac, and Jure Leskovec. "Learning Structural Node Embeddings via Diffusion Wavelets." KDD 2018. https://github.com/snap-stanford/graphwave

Ribeiro, Leonardo FR, Pedro HP Saverese, and Daniel R. Figueiredo. "struc2vec: Learning node representations from structural identity." KDD 2017. https://github.com/shenweichen/GraphEmbedding

Citation

If you find this useful, please use the following citation

@inproceedings{wang2021,
  title={Embedding Node Structural Role Identity Using Stress Majorization},
  author={Wang, Lili and Huang, Chenghan and Ma, Weicheng and Lu, Ying and Vosoughi, Soroush},
  booktitle={Proceedings of the 30th ACM International Conference on Information \& Knowledge Management},
  year={2021}
  
}