AI-Based Information Extraction from Collection Entry Records
Department of Greek, Etruscan and Roman Antiquities
France ✧ 2026
Artificial Intelligence Technologies Used
- OCR for Historical Documents
- Page layout analysis
- Named entities extraction
- Data Indexing and Structuring
Project Objectives
Extracting information from the Louvre's collection entry records
Extracted data :
Object description, dimensions, provenance information, material, price, inventory number, etc.
Processing workflow
- Corpus Analysis
- Drafting prompts (LLM) and testing phases using samples: one prompt for each structure (4 in total)
- Processing on all pages: data retrieved in JSON format
- Export to Excel format with the extracted data and metadata from the records
- Manual verification the consistency of the data found in each column
