LACA

Requirements

• python==3.9.18
• NetworkX==2.8.8
• Scipy==1.12.0
• Numpy==1.26.4
• sklearn==0.0.post12
• Third-party library pyrfm
• Third-party library LocalGraphClustering

Datasets

Get the processed data files from the corresponding .zip file and then, unzip and move files to attr_graphs/<dataset>.

Download processed data files

• cora
• pubmed
• blogcatalog
• flickr
• arxiv, reddit, yelp and amazon2m

Generate random seed nodes

You may use the randomly generated seed nodes used from our experiments to reproduce our result, or you may also generate them using attr_graphs/groundtruth.py.

Run LACA

python src/run.py --data cora --algo laca_c --alpha 0.9 --sigma 1 --dim 64 --e 1e-6
python src/run.py --data cora --algo laca_e --alpha 0.9 --sigma 1 --dim 32 --e 1e-6

Reproductivity

To reproduce our empirical results, run the run.sh script.