snudatalab / divmf

Diversely Regularized Matrix Factorization for Accurate and Aggregately Diversified Recommendation

This project is a pytorch implementation of Diversely Regularized Matrix Factorization for Accurate and Aggregately Diversified Recommendation. This paper proposes a novel approach, Diversely Regularized Matrix Factorization (DivMF), to achieve high aggregate level diversity in recommendation system while maintaining accuracy. This project provides executable source code with adjustable hyperparameters as arguments and preprocessed datasets which used in the paper.

Prerequisites

• Python 3.8+
• PyTorch
• NumPy
• Scipy
• Click
• tqdm

Usage

You can run a demo script demo.sh that reproduces the experimental results in the paper. You can change the hyperparameters by modifying arguments of main.py. Unpack zip files in data directory to use large datsets: Yelp-15, Gowalla-15, Movielens-10M.

Datasets

Preprocessed data are included in the data directory.

Name	Users	Items	Interactions	Download
Yelp-15	69,853	43,671	2,807,606	Link
Gowalla-15	34,688	63,729	2,438,708	Link
Epinions-15	5,531	4,286	186,995	Link
Movielens-10M	69.878	10,677	10,000,054	Link
Movielens-1M	6,040	3,706	1,000,209	Link

License

This software may be used only for non-commercial purposes (e.g., research evaluation) in universities. Please contact Prof. U Kang ([email protected]) if you want to use it for other purposes or use it in places other than universities.

Citation

@inproceedings{KimJLK23,
  author    = {Jongjin Kim and
               Hyunsik Jeon and
               Jaeri Lee and
               U Kang},
  title     = {Diversely Regularized Matrix Factorization for Accurate and Aggregately Diversified Recommendation},
  booktitle = {PAKDD},
  year      = {2023},
}