APGL4SR: A Generic Framework with Adaptive and Personalized Global Collaborative Information in Sequential Recommendation (CIKM'2023)

Source code for paper: APGL4SR: A Generic Framework with Adaptive and Personalized Global Collaborative Information in Sequential Recommendation (CIKM'2023)

Introduction

We incorporate adaptive and peronalized global collaborative information into sequential recommendation with the proposed APGL4SR framework.

Reference

If you find our article or implemented codes helpful, please kindly cite our work. Thank you!

@inproceedings{yin2023apgl4sr,
title={APGL4SR: A Generic Framework with Adaptive and Personalized Global Collaborative Information in Sequential Recommendation},
author={Yin, Mingjia and Wang, Hao and Xu, Xiang and Wu, Likang and Zhao, Sirui and Guo, Wei and Liu, Yong and Tang, Ruiming and Lian, Defu and Chen, Enhong},
booktitle={Proceedings of the 32nd ACM International Conference on Information and Knowledge Management},
pages={3009--3019},
year={2023}
}

Implementation

Requirements

Python >= 3.7
Pytorch >= 1.2.0
tqdm == 4.26.0 faiss-gpu == 1.7.1 nni == 2.10

Datasets

Four prepared datasets are included in data folder.

Train & Eval Model

cd src
chmod +x ./scripts/run_<DATASET>.sh
./scripts/run_<DATASET>.sh

where <DATASET> is the name of the four datasets.

Acknowledgment

• Transformer and training pipeline are implemented based on S3-Rec and ICLRec. Thanks to them for providing efficient implementation.