Making Provincial Archives Accessible: ML/AI-Driven Transformation of the Overijssel Resolutions (1578-1795)

NWO FUNDING: NWA.1518.22.105 Making Provincial Archives Accessible: ML/AI-Driven Transformation of the Overijssel Resolutions (1578-1795) C.A. Romein Acknowledgements: With help from Ben Wolf, Victor de Boer, Andreas Weber, Volunteers, Collectie Overijssel, HAICu-SCOPE. The “Resolutions” presented to Overijssel ●Direct evidence of citizen-government interaction ●Provincial citizenship in practice ●Rich source for spatial, temporal, and social analysis Spring 2028 Named Entity Recognition and Topics ●Layout labelling ●NER recognition ●Explore LLMs/tSNE/Annif to cluster content for topical metadata 2026-27 Recognizing the “Resolutions” ●Layout recognition for all sources. ●Text recognition for all sources. ●Start “finding meetings”. ●Hack-a-LOD: exploring maps for visualizations. ●Working with volunteers for Named Entity Recognition. 2025 HAICu @UT starts working on datafication of the “Resolutions” ●Exploring the layout and content. ●Testing models and tools. 2024 Digitisation of the “Resolutions” ●Provincial governance records to made accessible by 2028 (jubilee). ●Previously largely inaccessible for systematic research. ●Contains abstracts of requests/petitions of the inhabitants - these are hitherto unexplored due to their absence in indices. 2020 Meetings of the Staten van Overijssel ●118 bound manuscript volumes (1578-1795) ●~ 6,015 days of meeting-minutes ●~ 55,000+ handwrien pages ●~ 9,924,767 words in predominantly Dutch 1578-1795 This methodology provides provincial archives with a proven workflow for rendering large volumes of handwrien administrative sources accessible. The combination of automated text recognition, structured data (CIDOC-CRM, Wikidata), and citizen science validation is directly applicable to comparable resolution series in other provinces. Linking historical place names to Wikidata identifiers creates interoperability between heritage institutions, enabling researchers and the public to search sources across institutional boundaries. The project demonstrates how ML/AI technology augments rather than replaces archival expertise, with human interpretation remaining essential for quality assurance. This approach enables comparative research into early modern governance practices and connects scholarly investigation with public narratives about regional identity.