Federated Recommender System with Data Valuation for E-commerce Platform

This repository contains the official implementation of the paper "Federated Recommender System with Data Valuation for E-commerce Platform," published in Expert Systems with Applications (ESWA).
Paper (ESWA) | Preprint (arXiv)

Installation

1. Clone repository

git clone https://github.com/Kang-Min-Ku/FedGDVE.git

2. Download environment

   Download the pre-configured environment from the following link: 📦Drive

3. Download FedML module

   Since we modified a very old version of FedML, it needs to be downloaded from the following link manually: 📦Drive

4. Place FedML module in the root directory of this repository.

How to Run

1. Activate environment

   conda activate [absolute path of env]

2. Change directory

   cd experiments/distributed/recommendation

3. Run code

   sh run_fedgdve.sh [number of worker] [config file path]

   The configuration file is located in the config directory as a .yaml file.

Citation

@article{gdve2025federated,
  title={Federated recommender system with data valuation for e-commerce platform},
  author={Park, Jongwon and Kang, Minku and Sim, Wooseok and Lee, Soyoung and Park, Hogun},
  journal={Expert Systems with Applications},
  year={2025},
}