Cryptography, history and AI to decipher manuscripts

In 2022, Cécile Pierrot, a researcher in Inria's Caramba team (cryptography), had the idea of calling on Camille Desenclos, a specialist in 16th and 17th century cryptography, to successfully decipher a letter from Charles V that had resisted all attempts to decipher it until then.

"BACK IN TIME's ambitious project is to combine cryptography, history and artificial intelligence to unravel the mysteries of documents that have remained coded, sometimes for centuries. While the ultimate aim is to develop an automatic transcription and deciphering tool, this exploratory action should enable us to define the ways in which three disciplines, which do not spontaneously rub shoulders but have much to share, can work together."

The challenge is considerable, since not only have the languages evolved - some have even disappeared - but the authors have obviously not provided the encryption keys used.

"the difficulty for those seeking to decode is to manage to identify these correspondences and understand their logic, which requires a qualitative and quantitative classification of the signs used. "It is possible to manually transcribe two or three letters on a few pages, but when there are hundreds of letters, an automation tool is required.

Artificial intelligence is therefore called in, as part of the ALMAnaCH project, with the ultimate aim of producing a tool that will know how to use the right reading grid, whether for character recognition or deciphering.

This kind of multi-disciplinary collaboration is a formula favored by Inria, such is the fertile ground here.

For further information, click here:

BACK IN TIME : combining cryptography, history and AI to decipher manuscripts

Back In Time - Baliser l'Analyse CryptographiK de lettres anciennes grâce à l'INtelligence artifi-cielle : Technique et IMplémentation Exploratoire

Illustration: Henryk Niestrój - Pixabay

Visit inria.fr