This is the official implementation for the SIGIR 2024 paper "ReCODE: Modeling Repeat Consumption with Neural ODE" [PDF].
If you find our code or work useful for your research, please cite our work.
@inproceedings{dai2024recode,
title={ReCODE: Modeling Repeat Consumption with Neural ODE},
author={Dai, Sunhao and Qu, Changle and Chen, Sirui and Zhang, Xiao and Xu, Jun},
booktitle={Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval},
year={2024}
}
ReCODE is a novel model-agnostic framework that utilizes neural ODE to model repeat consumption. Moreover, ReCODE seamlessly integrates with various existing recommendation models, including collaborative-based and sequential-based models, making it easily applicable in different scenarios.
Please run the shell file "run.sh" to run our proposed ReCODE and other baselines.
./run.sh
Our experimental environment is shown below:
torch version: 1.8.0
- MMTD: The original data and description is avaliable here.
- Nowplaying-RS: The original data and description is avaliable here.
python main.py --gpu 0 --model_name MF --emb_size 32 --lr 5e-4 --l2 1e-7 --dataset "MMTD"
You can specify the gpu id, the model name, the used dataset by cmd line arguments.
Our implementations and experiments are conducted based on ReChorus benchmark. We use the Euler ODE solver from the torchdiffeq package.