Lorya – Unlocking Written Cultural Heritage with AI

Abstract

Lorya is an open, AI-powered platform that transforms digitised printed cultural heritage into clear, searchable digital text. It was developed by the UNDP Belgrade Office in partnership with the Mathematical Institute of the Serbian Academy of Sciences and Arts and the National Library of Serbia, with support from the governments of France and Japan. It is built on an advanced AI pipeline developed within the Republic of Serbia’s GovTech Programme 2023 to enhances the digitisation of old periodicals from the National Library of Serbia.

 Lorya provides the long-awaited solution to a challenge the National Library of Serbia faced for many years: achieving full-text searchability in its digital library—a breakthrough made possible only in the era of artificial intelligence. The platform integrates four AI-powered steps in the digitization workflow: image enhancement, layout identification, OCR, and post-OCR correction.

Version 1.0 of Lorya is designed for the Serbian language, with the National Library of Serbia as its first proud user. Future versions will be tailored for other, mostly digitally underrepresented languages.

Lorya in a nutshell:

Speaker