CSGNN

Codes of paper "Category-aware Self-supervised Graph Neural Network for Session-based Recommendation"

Requirements

• Python 3
• PyTorch 1.6.0

you can create a new environment by conda

conda install --yes --file requirements.txt

Usage

datasets

1. directory /datasets/id/ is the dataset of CSGNN model
2. directory /datasets/pre_training/ is the embeddings of pre-training
3. directories /datasets/filter* are the datasets of ablation experiments

Train and evaluate

python main.py --dataset nowplaying --beta 0.005 --embSize 100 > result.output