STEP-BACK PROFILING: Distilling User History for Personalized Scientific Writing

This repository contains the code and dataset for the paper "STEP-BACK PROFILING: Distilling User History for Personalized Scientific Writing".

Table of Contents

• Dataset Generation
• Results Generation
• LaMP Result
• Citation
• License

Dataset Generation

The dataset generation process involves the following steps:

0. Download the raw data here s2orc_4000.json

1. Get sampled author list and paper list in JSON format:

   • dataset/data_construction.ipynb

2. Extract author's research interests:

   • dataset/s2orc-rq.ipynb

3. Extract research questions from papers:

   • dataset/research_question_extraction.ipynb

PSW Results Generation

For generating results for each task, follow these steps:

1. Get User Profile:

   • psw_result/author_profiling_cot.ipynb

2. Generate title for single author:

   • psw_result/single_agent_title_generation.ipynb

3. Generate results for multiple authors & evaluation for each task:

   • psw_result/task1_solving.ipynb
   • psw_result/task2_solving.ipynb
   • psw_result/task3_solving.ipynb
   • psw_result/task4_solving.ipynb

LaMP Results Generation

The lamp_result/ directory contains the following notebooks:

• lamp_result/cot_generation.ipynb
• lamp_result/final_output_generation.ipynb
• lamp_result/user_profile_generation.ipynb

These notebooks are used for generating user profiles and final outputs for the LaMP dataset.

Citation

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

@misc{tang2024stepback,
      title={Step-Back Profiling: Distilling User History for Personalized Scientific Writing}, 
      author={Xiangru Tang and Xingyao Zhang and Yanjun Shao and Jie Wu and Yilun Zhao and Arman Cohan and Ming Gong and Dongmei Zhang and Mark Gerstein},
      year={2024},
      eprint={2406.14275},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
}

License

This project is licensed under the MIT License. See the LICENSE file for more details.