ECoC

We study the problem of sequential recommendation based on reinforcement learning. More specifically, we propose an Efficient Continuous Control (ECoC) framework to facilitate unified action learning, under the framework of actor-critic.

Access the dataset

Download the preprocessed datasets Tmall and Yelp.

How to run our code

1. unzip the data, put the data folder in the data/ directory
2. run the corresponding script within runs/

Main Packages

torch           
numpy  
pandas
tensorboard
tqdm

Citation

If you use our code, please cite the paper

@article{wang2024efficient,
  title={An Efficient Continuous Control Perspective for Reinforcement-Learning-based Sequential Recommendation},
  author={Wang, Jun and Wu, Likang and Liu, Qi and Yang, Yu},
  journal={arXiv preprint arXiv:2408.08047},
  year={2024}
}