RFCRationaleBuilder

RFCRationaleBuilder is the codebase for our curated dataset CodeConvo, available at https://huggingface.co/datasets/jiebi/CodeConvo/. CodeConvo includes software engineering repositories (e.g., nixpkgs, kubernetes, pytorch, react, rust, freecodecamp, vscode), documentation repositories such as kubernetes-website, plus SWE-bench (converted to our formats), and many IETF WG Internet-Drafts repositories.

Two dataset variants are produced:

• ids: PR/edit ↔ issue/comment relationships from IETF draft evolution and GitHub discussions.
• code: PR/code-change ↔ issue/comment relationships from software iteration and discussions.

Setup Instructions

Prerequisites

• Python 3.8 or higher
• For pygit2 on macOS: brew install libgit2
• For pygit2 on Ubuntu/Debian: sudo apt-get install libgit2-dev

1. Create a Python Virtual Environment

# Create virtual environment
python3 -m venv venv

# Activate virtual environment
# On macOS/Linux:
source venv/bin/activate

# On Windows:
venv\Scripts\activate

2. Install Required Packages

# Install all dependencies from requirements.txt
pip install -r requirements.txt

3. Verify Installation

# Check that packages are installed
pip list

Note: Using the Virtual Environment

Always activate the virtual environment before running scripts:

# Activate (macOS/Linux)
source venv/bin/activate

# Deactivate when done
deactivate

Processing Pipeline

The full pipeline (with edge cases) is documented in scripts/README.md.

Dataset Citation

If you use CodeConvo or this pipeline, please cite:

@inproceedings{bian2024tell,
  title={Tell Me Why: Language Models Help Explain the Rationale Behind Internet Protocol Design},
  author={Bian, Jie and Welzl, Michael and Kutuzov, Andrey and Arefyev, Nikolay},
  booktitle={2024 IEEE International Conference on Machine Learning for Communication and Networking (ICMLCN)},
  pages={447--453},
  year={2024},
  organization={IEEE}
}

@article{bian2025automated,
  title={Automated insights into github collaboration dynamics},
  author={Bian, Jie and Arefev, Nikolay and M{\"u}hlh{\"a}user, Max and Welzl, Michael},
  journal={IEEE Access},
  year={2025},
  publisher={IEEE}
}

Dataset Usage

Two versions are available:

• ids variant (binary classification): Focuses on Internet-Drafts with natural "paragraphs" (see scripts/txt2csv_ids.py). Download: https://huggingface.co/datasets/jiebi/ids-paragraph-commentary-corpus
• code variant (information retrieval): Generalized to git repositories (see scripts/txt2csv_code.py). Download: https://huggingface.co/datasets/jiebi/CodeConvo

Presentation

See poster presented at ICMLCN 2024: https://github.com/cheop-byeon/bert_for_long_text/blob/main/img/ICMLCN2024_poster.pdf (conference program)

Acknowledgements

• gh2md
• xml2rfc
• mmark
• kramdown-rfc