LOAM

This is official Pytorch implementation of our paper LOAM: Improving Long-tail Session-based Recommendation via Niche Walk Augmentation and Tail Session Mixup, accepted by SIGIR '23

Datasets

Datasets we used in the paper can be downloaded from:

https://drive.google.com/drive/folders/1UnrR6w6dRQhnCIRCq0voN58cwwX5rHDV?usp=share_link

Unzip the datasets and move them to Datasets/.
You can also preprocess raw datasets downloaded from public links by running .ipynb files in Datasets/preprocess_code.

Requirements

• Python 3
• PyTorch
• NetworkX
• Numpy
• wandb

Basic Usage

• Change the experimetal settings and model hyperparameters using the config.py
• Run main.py --dataset [dataset_name] to train and test models.
• You can record performance and loss values by setting wandb.

Reference

• SRGNN
• NISER
• CORE

Citation

Please cite our paper if you use the code: