psm1206 / MAWU

[CIKM'23] "Toward a Better Understanding of Loss Functions for Collaborative Filtering"

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Toward a Better Understanding of Loss Functions for Collaborative Filtering (CIKM'23)

This is the official code for MAWU in the paper "Toward a Better Understanding of Loss Functions for Collaborative Filtering", The 32nd ACM International Conference on Information and Knowledge Management. This code is implemented on RecBole.

Note that a summary of our paper is on our lab blog (in Korean).

How to run

Set conda environment

conda env create -f mawu.yaml
conda activate mawu

Run commands for DirectAU

python run_recbole.py --model=DirectAU --dataset=beauty --encoder=MF --weight_decay=1e-4 --gamma=0.4 &&
python run_recbole.py --model=DirectAU --dataset=beauty --encoder=LightGCN --weight_decay=1e-4 --gamma=0.4 &&
python run_recbole.py --model=DirectAU --dataset=gowalla --encoder=MF --weight_decay=1e-6 --gamma=2 &&
python run_recbole.py --model=DirectAU --dataset=gowalla --encoder=LightGCN --weight_decay=1e-6 --gamma=2 &&
python run_recbole.py --model=DirectAU --dataset=yelp --encoder=MF --weight_decay=1e-6 --gamma=2 &&
python run_recbole.py --model=DirectAU --dataset=yelp --encoder=LightGCN --weight_decay=1e-6 --gamma=2

Run commands for MAWU

python run_recbole.py --model=MAWU --dataset=beauty --encoder=MF --weight_decay=1e-4 --gamma1=1 --gamma2=0.1 &&
python run_recbole.py --model=MAWU --dataset=beauty --encoder=LightGCN --weight_decay=1e-4 --gamma1=0.9 --gamma2=0.2 &&
python run_recbole.py --model=MAWU --dataset=gowalla --encoder=MF --weight_decay=1e-6 --gamma1=2.6 --gamma2=1.4 &&
python run_recbole.py --model=MAWU --dataset=gowalla --encoder=LightGCN --weight_decay=1e-6 --gamma1=2.4 --gamma2=1.6 &&
python run_recbole.py --model=MAWU --dataset=yelp --encoder=MF --weight_decay=1e-6 --gamma1=0.8 --gamma2=0.6 &&
python run_recbole.py --model=MAWU --dataset=yelp --encoder=LightGCN --weight_decay=1e-6 --gamma1=1.2 --gamma2=0.6

Citation

If you find our work helpful, please cite our paper.

@inproceedings{park2023mawu,
  title={Toward a Better Understanding of Loss Functions for Collaborative Filtering},
  author={Seongmin Park and
          Mincheol Yoon and
          Jae-woong Lee and
          Hogun Park and
          Jongwuk Lee},
  booktitle={The 32nd ACM International Conference on Information and Knowledge Management (CIKM)},
  year={2023}
}

About

[CIKM'23] "Toward a Better Understanding of Loss Functions for Collaborative Filtering"

License:MIT License


Languages

Language:Python 100.0%