Bipartite Graph Clustering

Requirements

• Linux machine
• Python3
• numPy = 1.23.3
• scikit-learn = 1.1.2
• scipy = 1.9.2
• argparse = 1.1
• munkres = 1.1.4

Large Datasets

Please download them from link

Clustering and Evaluation

$ sh evaluate.sh BGC cora
$ sh evaluate.sh FNEM cora
$ sh evaluate.sh SNEM cora

or

$ python3 -W ignore run.py --algo BGC --data cora --alpha 0.3 --dim 5
$ python3 -W ignore eval.py --algo BGC --data cora
$
$ python3 -W ignore run.py --algo FNEM --data cora --alpha 0.3 --dim 5
$ python3 -W ignore eval.py --algo FNEM --data cora
$
$ python3 -W ignore run.py --algo SNEM --data cora --alpha 0.3 --dim 5
$ python3 -W ignore eval.py --algo SNEM --data cora

Note that "--dim 5" means that the beta parameter in the paper is set to 5*k. The clustering results can be found in the folder "cluster".

Citation

@article{YangShi23,
  author       = {Renchi Yang and
                  Jieming Shi},
  title        = {Efficient High-Quality Clustering for Large Bipartite Graphs},
  journal      = {Proceedings of the ACM on Management of Data},
  volume       = {2},
  number       = {1},
  pages        = {23:1--23:27},
  year         = {2024}
}