This is the PyTorch implementation for our paper "FairCDR: Transferring Fairness and User Preferences for Cross-Domain Recommendation."
The dataset used can be found at: https://tianchi.aliyun.com/dataset/408.
The data preprocessing method is described in detail here: https://github.com/datawhalechina/torch-rechub/tree/main/examples/ranking/data/ali-ccp.
The dataset is divided into three domains (1, 2, 3) based on the "Context Features."
- python==3.8
- pytorch>=1.10.0
- numpy>=1.17.2
- scipy>=1.6.0
To obtain user and item representations for the target and source domains, run the following command:
python main.py --train_type="pretrain" To perform fairness and user preferences transfer for Cross-Domain Recommendation, run the following command:
python main.py --train_type="train"