Reproducible artifact and evaluation package for the research paper on AI-assisted facilitation in high-stakes workshops.

Paper: [arXiv link - to be added after submission]

• facilcopilot/ - Core Python modules
• data/scenarios/ - Replay evaluation scenarios (D2 dataset)
• data/examples/ - Example transcript schema
• experiments/configs/ - Experiment configuration files
• experiments/runs/ - Evaluation results (for verification)
• scripts/ - Evaluation and analysis scripts
• tests/ - Unit and integration tests

# Install dependencies
pip install -e .

# Run tests
make test

# Run evaluation pipelines
make eval_detection
make eval_replay

# Generate paper tables
make build_paper_tables

All experiments use fixed random seeds. Configs, hyperparameters, and scenario definitions are version-controlled. The paper (available on arXiv) documents all testable claims and evidence requirements. Evaluation results in experiments/runs/ can be reproduced by running the scripts in scripts/ with the provided configuration files.

If you use this code or data, please cite our paper:

[Citation will be added after arXiv publication]

[To be determined]