ResScope

The open source code for analysis tools mentioned in "Ads Recommendation in a Collapsed and Entangled World " presented at SIGKDD 2024. We dive into the analysis tools and results regarding several fundamental aspects of recommendation models: the dimensional collapse of embeddings, the correlation between features, and the entanglement of user interests.

🚧 This code repository is under construction, stay tuned!

• The analysis of dimensional collapse is from the code for our ICML 2024 paper: On the Embedding Collapse when Scaling up Recommendation Models. arXiv, code
• The analysis of feature correlation is from the code for our WWW 2024 paper: Temporal Interest Network for User Response Prediction. arXiv, code
• The analysis of interest entanglement is from the code for our AAAI 2024 paper: STEM: Unleashing the Power of Embeddings for Multi-task Recommendation. arXiv, code

Contact

For questions regarding the methods themselves, feel free to create an issue or send an email to

• [email protected]

Citation

If you find our code helpful in your research, please kindly cite the following papers:

@article{pan2024ad,
  title={Ad Recommendation in a Collapsed and Entangled World},
  author={Pan, Junwei and Xue, Wei and Wang, Ximei and Yu, Haibin and Liu, Xun and Quan, Shijie and Qiu, Xueming and Liu, Dapeng and Xiao, Lei and Jiang, Jie},
  journal={SIGKDD},
  year={2024}
}

@article{guo2023embedding,
  title={On the Embedding Collapse when Scaling up Recommendation Models},
  author={Guo, Xingzhuo and Pan, Junwei and Wang, Ximei and Chen, Baixu and Jiang, Jie and Long, Mingsheng},
  journal={ICML},
  year={2023}
}

@inproceedings{zhou2024temporal,
  title={Temporal Interest Network for User Response Prediction},
  author={Zhou, Haolin and Pan, Junwei and Zhou, Xinyi and Chen, Xihua and Jiang, Jie and Gao, Xiaofeng and Chen, Guihai},
  booktitle={Companion Proceedings of the ACM on Web Conference 2024},
  pages={413--422},
  year={2024}
}

@inproceedings{su2024stem,
  title={STEM: Unleashing the Power of Embeddings for Multi-task Recommendation},
  author={Su, Liangcai and Pan, Junwei and Wang, Ximei and Xiao, Xi and Quan, Shijie and Chen, Xihua and Jiang, Jie},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={38},
  number={8},
  pages={9002--9010},
  year={2024}
}

Contributors

We gratefully acknowledge the contributions of the following: Xingzhuo Guo, Yu Kang, Mingjia Yin, Zhutian Lin, Jiancheng Wang, Haolin Zhou and Xinyi Zhou.