/data: benchmark datasets
/log: storage of results
model.py: define a two-layer GCN backbone model
sag.py: train and evaluate SAG

python == 3.7.10
pytorch == 1.8.0
CUDA == 11.4
torch-geometric == 2.1.0
torch-scatter == 2.0.6
torch-sparse == 0.6.12
scipy == 1.5.2
numpy == 1.19.5
scikit-learn == 0.22.1

python sag.py

python sag.py --dataset Citeseer  --lamb 0.5 --theta 0.01
python sag.py --dataset Pubmed --lamb 0.2 --theta 0.01

@inproceedings{yang2023mitigating,
  title={Mitigating Semantic Confusion from Hostile Neighborhood for Graph Active Learning},
  author={Yang, Tianmeng and Zhou, Min and Wang, Yujing and Lin, Zhengjie and Pan, Lujia and Cui, Bin and Tong, Yunhai},
  booktitle={Proceedings of the 32nd ACM International Conference on Information and Knowledge Management},
  pages={4380--4384},
  year={2023}
}