Neural Graph Collaborative Filtering with MovieLens in torch

Dataset

This repository is about Neural Graph Collaborative Filtering with MovieLens in torch. Dataset is Implict Feedback, If there is interaction between user and item, then target value will be 1. So if there is rating value between user and movie, then target value is 1, otherwise 0. For negative sampling, ratio between positive feedback and negative feedback is 1:4 in trainset, and 1:99 in testset. (these ratios are same as NCF setting)

Result

I measured NDCG@10 and HitRatio@10 while changing the number of embedding layers for MovieLens dataset 100k and 1M.

dataset	Best NDCG@10	HR@10	# layers	epoch	batch size
MovieLens100k	0.5784	0.8164	3	20	256
MovieLens100k	0.5640	0.8262	4	20	256
MovieLens100k	0.5546	0.8377	5	20	256
MovieLens1m	0.4964	0.7568	3	20	256
MovieLens1m	0.4922	0.7614	4	20	256
MovieLens1m	0.4849	0.7507	5	20	256

Dependency

pytorch >= 1.12.0
python >= 3.8
scipy >= 1.7.1
numpy >= 1.20.3

Quick Start

python3 main.py -e 10 -b 256 -dl true -k 10 -fi 100k

Reference

1. Neural Graph Collaborative Filtering

2. Official code from Xiang Wang

   @inproceedings{NGCF19,
     author    = {Xiang Wang and
                  Xiangnan He and
                  Meng Wang and
                  Fuli Feng and
                  Tat{-}Seng Chua},
     title     = {Neural Graph Collaborative Filtering},
     booktitle = {Proceedings of the 42nd International {ACM} {SIGIR} Conference on
                  Research and Development in Information Retrieval, {SIGIR} 2019, Paris,
                  France, July 21-25, 2019.},
     pages     = {165--174},
     year      = {2019},
   }