Intrinsic Extrinsic Disentangled Recommendation (IEDR)

This is the source code for the Intrinsic Extrinsic Disentangled Recommendation (IEDR) model proposed in the paper "Intrinsic and Extrinsic Factor Disentanglement for Recommendation in Various Context Scenarios" published on Transactions on Information Systems (TOIS).

The instructions to create the environment for IEDR.

We run the code on cuda version 11.3 or 11.4, or cpu.

1. Create a conda environment

conda create -n py39_torch python=3.9 scipy numpy
conda activate py39_torch

2. Install the pytorch 11.2.0

# GPU version
conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch

# CPU version
conda install pytorch torchvision torchaudio cpuonly -c pytorch

3. Install torch_geometric

# GPU version
CUDA=cu113
TORCH=1.12.0

# CPU version
CUDA=cpu
TORCH=1.12.0

pip install torch-scatter -f https://data.pyg.org/whl/torch-${TORCH}+${CUDA}.html
pip install torch-sparse -f https://data.pyg.org/whl/torch-${TORCH}+${CUDA}.html
pip install torch-geometric

4. Install other packages

pip install tensorboard
pip install icecream
pip install numpy
pip install pandas

Run the Code

cd run
bash frappe_run.sh

Cite our paper

Please credit our work by citing the following paper:

@article{su2025intrinsic,
  title={Intrinsic and Extrinsic Factor Disentanglement for Recommendation in Various Context Scenarios},
  author={Su, Yixin and Jiang, Wei and Lin, Fangquan and Yang, Cheng and Erfani, Sarah Monazam and Gan, Junhao and Zhao, Yunxiang and Li, Ruixuan and Zhang, Rui},
  journal={Transactions on Information Systems (TOIS)},
  year={2025}
}