The HAICu (digital Humanities, Artificial Intelligence, Cultural heritage) consortium is a Dutch national research effort, funded under the Nationale Wetenschapsagenda, working at the intersection of AI, the humanities, and cultural-heritage practice. My contribution sits within Workpackage 2, Continual Machine Learning with Humans in the Loop, hosted at the University of Twente (BMS-KiTeS).
The empirical focus is the Resolutieboeken of the provincial States of Overijssel — the principal record of provincial governance in the seventeenth and eighteenth centuries, running effectively from 1528/1578 to 1795. The aim is twofold. The first is technical: to develop and evaluate continual-learning approaches to handwritten text recognition for a corpus that spans nearly three centuries of changing handwriting conventions. The second is methodological and infrastructural: to make the resulting transcriptions, segmentations, and metadata genuinely usable for historians and for the regional archives that hold the source.
I co-coordinate the Histories of Ordinary People Innovation Lab within HAICu (with René Duursma, Groninger Archieven), which is the connective tissue between the academic research and the GLAM partners.