/MCL

Code for the WWW'22 paper "MCL: Mixed-Centric Loss for Collaborative Filtering"

Primary LanguagePython

WWW'22 MCL: Mixed-Centric Loss for Collaborative Filtering

Authors: Zhaolin Gao*, Zhaoyue Cheng*, Felipe Perez, Jianing Sun, Maksims Volkovs [paper]

Environment:

The code was developed and tested on the following python environment:

python 3.7.7
pytorch 1.9.0
scikit-learn 0.23.2
numpy 1.19.1
scipy 1.5.4
tqdm 4.48.2
tensorboard 2.7.0

Instructions:

Train and evaluation LightGCN + MCL:

Training on Amazon-Digital-Music dataset:

python main.py --dataset amazon-digital-music --alpha 1.25 --beta 5 --lamb_p 6.5 --lamb_n -0.5

Training on Amazon-Grocery dataset:

python main.py --dataset amazon-grocery --alpha 1.25 --beta 5 --lamb_p 6.5 --lamb_n -0.5

Training on Amazon-Books dataset:

python main.py --dataset amazon-book --alpha 1 --beta 4 --lamb_p 8 --lamb_n -1

Training on Yelp2021 dataset:

python main.py --dataset yelp --alpha 1 --beta 4 --lamb_p 8 --lamb_n -1

Citation

If you find this code useful in your research, please cite the following paper:

@inproceedings{gao2022mcl,
  title={MCL: Mixed-Centric Loss for Collaborative Filtering},
  author={Zhaolin Gao, Zhaoyue Cheng, Felipe Perez, Jianing Sun, Maksims Volkovs},
  booktitle={Proceedings of the International World Wide Web Conference},
  year={2022}
}