Awesome

SEAL

Codes and data for KDD2020 submission.

Usage

Un-tar processed datasets:

chmod +x untar_files.sh
./untar_files.sh

Or you could run pre-processing scripts in preprocessing/ folder to obtain datasets from raw data.

Run the algorithm:

python run_seal_attr.py --dataset dblp --locator_coef 0.5 --radius_penalty 0.0

Dependencies

All dependencies are included in env.yml.

You could install all dependencies with conda:

conda env create -f env.yml

Datasets

Communities in *.cmty.txt have been randomly shuffled. So the first 450 are training communities, the following 50 are validation communities, and the remaining is the test set.