CodeBook Studio
CodeBook Studio is a codebook-driven text annotation tool for computational social science. It lets researchers define an annotation task, annotate texts, export the task as codebook.json, and save the human-annotated data as ground-truth.csv for downstream LLM evaluation.
It is designed to reduce the overhead of building annotation workflows from scratch. Researchers can define a task once, use it for human annotation, preview the corresponding LLM prompts, and carry the same codebook plus the human-coded ground-truth.csv directly into CodeBook Lab for model-based experiments.
CodeBook Studio provides the following features:
- Consistent experience for teams of annotators
- Multiple annotation types: binary, categorical, Likert-scale, and open-ended text
- Fully customisable codebook editor with integrated instructions and examples
- Codebook export in LaTeX and Markdown formats for easy rendering in papers and appendices
- LLM prompt preview and export for pipeline integration
- CSV input and output for interoperability
The app is available at codebook.streamlit.app. Feel free to reach out if you encounter any bugs or need additional features for your annotation workflow.