MetaMate

AI-assisted data extraction for educational systematic reviews and meta-analyses.

Why MetaMate?

Data extraction is one of the most time-consuming steps in a systematic review or meta-analysis — manually locating and coding sample sizes, study designs, and intervention details across dozens of papers.

MetaMate uses AI to generate a first draft of your extraction, then gives you the tools to verify every value: confidence scores, evidence type indicators, page citations, and a synchronized PDF viewer — all in one interface. Define your own coding scheme, verify the results, and export to CSV. No programming required.

Features

• Researcher-defined coding schemes — not limited to PICO; supports any variables relevant to your review
• Confidence scores and evidence type indicators (Direct Evidence vs. Inference) for each extracted value
• Synchronized PDF viewer for side-by-side verification of extracted data
• Page-level citations linking each value back to the source text
• CSV export for use with your statistical analysis software
• Batch processing across multiple PDFs

Getting Started

No installation needed — just open the link above and start extracting.

Run locally with Docker

1. Install Docker

Download and install Docker Desktop and make sure it's running.

2. Get an LLM API key

You need an API key from an LLM provider. OpenAI is the default.

3. Download and run MetaMate

Open a terminal (Mac: Terminal app, Windows: PowerShell) and run:

git clone https://github.com/GaoxiangLuo/OpenMetaMate.git
cd OpenMetaMate
cp .env.example .env

Open the .env file in a text editor and replace sk-proj-your-api-key-here with your API key from step 2. Then start the application:

docker-compose up --build

Once you see the services are running, open http://localhost:3000 in your browser.

To stop, press Ctrl+C in the terminal, then run:

docker-compose down

Supported LLM providers: OpenAI, Google (Gemini), OpenRouter, or any OpenAI-compatible API (vLLM, Ollama, etc.). See .env.example for all configuration options.

Manual setup (without Docker)

# Backend
cd backend
pip install uv && uv sync
uv run uvicorn app.main:app --reload --port 8000

# Frontend (in a new terminal)
cd frontend
pnpm install && pnpm dev

Citation

If you use MetaMate in your research, please cite:

@inproceedings{10.1145/3772363.3798755,
  author    = {Wang, Xue and Luo, Gaoxiang},
  title     = {MetaMate: Understanding How Educational Researchers Experience AI-Assisted Data Extraction for Systematic Reviews},
  year      = {2026},
  publisher = {Association for Computing Machinery},
  address   = {New York, NY, USA},
  url       = {https://doi.org/10.1145/3772363.3798755},
  doi       = {10.1145/3772363.3798755},
  booktitle = {Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems},
  series    = {CHI EA '26}
}

Documentation

Document	Description
Contributing Guide	How to contribute, code standards, PR process
Cloud Deployment	Self-hosting on AWS with Terraform
Infrastructure	Terraform configuration reference

Project Structure

OpenMetaMate/
├── backend/               # FastAPI backend
│   ├── app/
│   │   ├── api/routes/    # API endpoints
│   │   ├── core/          # Configuration & exceptions
│   │   ├── models/        # Pydantic schemas
│   │   └── services/      # LLM, PDF processing, S3
│   ├── pyproject.toml
│   └── Dockerfile
├── frontend/              # Next.js frontend
│   ├── app/               # App Router pages
│   ├── components/        # React components
│   ├── lib/               # API client, types, utilities
│   └── Dockerfile
├── infra/                 # Terraform IaC
├── docs/                  # Extended documentation
├── plans/                 # Architecture decision records
├── docker-compose.yml
└── .env.example

Contributing

We welcome contributions! See our Contributing Guide for the development workflow, code standards, and PR process.

License

CC BY-NC 4.0 — see LICENSE