This is the official pytorch implementation of RecSys'25 paper "IP2: Entity-Guided Interest Probing for Personalized News Recommendation".

If you use pip, please use requirements.txt to install the required packages. If you use conda, please use ip2_env.yaml to create the environment. Please make sure your GPU driver is correctly installed.

Our default environment is Debian 12 with pytorch 2.0.1+cu117.

Due to the copyright issue, we cannot include data in our repository. Please refer to the README.md file in the data folder for more details on data preparation.

Please carefully check the config.yaml file to make sure key parameters, data paths, and model parameters are correctly set. Once everything is ready, you can run the code with:

python main.py

In the first run, it will take a little bit longer time to process the data. You can use tensorboard to monitor the training process and the testing results.

If you find this repository useful, please cite our paper:

@inproceedings{wu2025IP2,
    title={IP2: Entity-Guided Interest Probing for Personalized News Recommendation},
    author={Wu, Youlin and Sun, Yuanyuan and Zhang, Xiaokun and Zhan, Haoxi and Xu, Bo and Yang, Liang and Lin, Hongfei},
    booktitle={Proceedings of the 19th ACM conference on recommender systems},
    year={2025},
    doi={10.1145/3705328.3748091}
}