Digital application, 2024-2025

UrbIA

We were asked to develop an AI-tool to help new generations of civil servants to handle a particularly complex data set for everyday purposes.

The historical land registry books from the 19th and 20th centuries are handwritten records, structured within pre-printed tables. Their format presents a significant challenge for modern handwritten text recognition technologies.

To address this, we conducted a comprehensive audit of the books, analysing their structure and how they have been used for administrative purposes. Based on these findings, we developed an AI-driven transcription model in collaboration with LITIS of the Rouen University.

The project was commissioned by the Administration du Cadastre et de la Topographie and supported by the Luxembourg Ministry for Digitalisation.