PyTorch Implementation for Neural Graph Collaborative Filtering

This is a PyTorch Implemenation for this paper:

Xiang Wang, Xiangnan He, Meng Wang, Fuli Feng, and Tat-Seng Chua (2019). Neural Graph Collaborative Filtering. SIGIR 2019

Original TensorFlow Implementation can be found here

Environment Requirement

You may simply run pip install -r requirements.txt

In this implementation, we use Python 3.7.5 with CUDA 10.1. The required packages are as follows:

• pytorch==1.3.1
• numpy==1.16.4
• scipy==1.4.1
• scikit-learn==0.22

Run the Code

NGCF

• Gowalla:

python main.py --dataset gowalla --alg_type ngcf --regs [1e-5] --embed_size 64 --layer_size [64,64,64] --lr 0.0001 --save_flag 1 --pretrain 0 --batch_size 4096 --epoch 500 --verbose 1 --mess_dropout [0.1,0.1,0.1] 

• Amazon-book:

python main.py --dataset amazon-book --alg_type ngcf --regs [1e-5] --embed_size 64 --layer_size [64,64,64] --lr 0.0005 --save_flag 1 --pretrain 0 --batch_size 4096 --epoch 500 --verbose 1 --mess_dropout [0.1,0.1,0.1]

MF

• Gowalla:

python main.py --dataset gowalla --alg_type mf --regs [1e-5] --embed_size 64  --lr 0.0001 --save_flag 1 --pretrain 0 --batch_size 4096 --epoch 500 --verbose 1 

• Amazon-book:

python main.py --dataset amazon-book --alg_type mf --regs [1e-5] --embed_size 64  --lr 0.0005 --save_flag 1 --pretrain 0 --batch_size 4096 --epoch 500 --verbose 1 --mess_dropout [0.1,0.1,0.1]

Dataset

Datasets and Data files are the same as thoese in the original repository.