SorbonneUniversity/LAD

Multilingual Museum Artifact Extractor

A Streamlit application for extracting artifact information from multilingual museum catalogs and aligning artifact names across English, Arabic, and French.

Hugging Face Space deployment

This repository is configured for a Hugging Face Docker Space. The container starts the Streamlit app with:

streamlit run app.py --server.address=0.0.0.0 --server.port=7860

Before starting the Space, add the required secrets in the Space settings.

Required, depending on the models you choose in the UI:

• —OPENAI_API_KEY
• —MISTRAL_API_KEY
• —GOOGLE_API_KEY

Optional, for database save/cache features:

• —ENABLE_SUPABASE=true
• —SUPABASE_URL
• —SUPABASE_ANON_KEY

Features

• —Multilingual document processing for English, Arabic, and French
• —OCR with adaptive correction
• —Artifact metadata extraction and cross-language alignment
• —Review, export, and optional Supabase persistence
• —Streamlit UI with per-run logs and provenance views

Project structure

• —app.py: Streamlit entrypoint
• —main.py: CLI entrypoint
• —config.py: configuration defaults
• —modules/: processing, OCR, extraction, validation, and database helpers
• —prompts/: prompt templates
• —requirements.txt: Python dependencies
• —Dockerfile: Hugging Face Space container runtime