Transcribing Handwritten Documents with Artificial Intelligence
Year after year, for over twenty years, BAnQ has enriched its digital offerings thanks to its digitization program. This presentation identifies partnership with a huge potential to facilitate access to digital content. Funded by SSHRC and designed at Université de Québec à Rimouski, the research project “Nouvelle-France numérique” aims to better document the mechanisms of documentary production and its networks of actors in New France. This project notably builds on the computing platform READ (Recognition and Enrichment of Archival Documents), which uses artificial intelligence to develop automatic recognition of handwriting. Using models prepared by researchers or archivists familiar with sources from the French Regime, the tool manages to transcribe the documents submitted for processing with an increasing and already very high efficiency. Over the course of the next two years, BAnQ will contribute to this partnership by sharing the knowledge of its staff on the fonds and documents of the archives of New France and by making available to researchers the sources to be modeled and processed by the transcription software. Produced in a standardized and expandable format (XML-TEI), the transcriptions could be reused and enriched in the framework of future research. They will also be integrated into BAnQ’s research tools for the benefit of its users.