The goal of this project is to model user interest and conformity in a time-aware manner. TDIC disentangles interest and conformity using item popularity.
- Python = 3.10.14
- Pytorch = 2.1.0
- numpy = 1.26.4
- tqdm = 4.64.4
Train and evaluate the model with the following commands. You can also add command parameters to specify dataset/epoch, or set neg-sampling rate.
python src/run.py --config src/config/xyz.cfgIf you use our codes in your research, please cite:
@INPROCEEDINGS{11169623,
author={Li, Qibo and Zhao, Yuqi and Ma, Yutao},
booktitle={2025 IEEE International Conference on Web Services (ICWS)},
title={TDIC: Time-Aware Disentanglement of Interest and Conformity in Mobile App Recommendations},
year={2025},
pages={1-10},
doi={10.1109/ICWS67624.2025.00011}}