ITEM is a framework to train MP-GNN with ranking losses (SmoothNDCG,NeuralNDGC, SmoothAP...) Accepted at TMLR.
conda create -n item python=3.9
conda activate item
pip install -e .
Then install package in requirements.txt
BPR Loss
sh scripts/lgcn_bpr_yelp.sh
SmoothNDCG Loss
sh scripts/lgcn_item_yelp.sh
@misc{karmim2024itemimprovingtrainingevaluation,
title={ITEM: Improving Training and Evaluation of Message-Passing based GNNs for top-k recommendation},
author={Yannis Karmim and Elias Ramzi and Raphaël Fournier-S'niehotta and Nicolas Thome},
year={2024},
eprint={2407.07912},
archivePrefix={arXiv},
primaryClass={cs.IR},
url={https://arxiv.org/abs/2407.07912},
}
