Do Loyal Users Enjoy Better Recommendations?: Understanding Recommender Accuracy from a Time Perspective

Code to reproduce the experiments from the paper: Do Loyal Users Enjoy Better Recommendations?: Understanding Recommender Accuracy from a Time Perspective (ICTIR2022)

This repository has the implementations for five models:

1. BPR
2. NeuMF: we follow https://github.com/hexiangnan/neural_collaborative_filtering
3. LightGCN: we follow https://github.com/RUCAIBox/RecBole
4. SASRec: we follow https://github.com/pmixer/SASRec.pytorch
5. TiSASRec: we follow https://github.com/JiachengLi1995/TiSASRec

Dataset

Dataset can be downloaded from https://drive.google.com/drive/folders/1TyQysstuaUo4IYb3WaVt4dVFOyZu3oN_.

Environment Requirement

BPR

• Tensorflow 1.14
• Python 3.6.9

NeuMF

• Tensorflow 1.14
• Python 3.6.9
• Keras 2.3.0

LightGCN

• Install RecBole package in https://github.com/RUCAIBox/RecBole

SASRec

• PyTorch >= 1.6

TiSASrec

• Tensorflow 1.14

Data formart

User ID	Item Id	Rating	Timestamp	Year
...	...	...	...	...

Examples to run the code

cd BPR/
python test.py --path ../Dataset/ --data movielens --selected_year 5 --gpu 1