Nowadays, Digital Humanities are challenged by the growing prominence of computational and AI-based approaches, sometimes at the expense of data structuration. This article argues that rigorous data modelling remains a central epistemological component of Digital Humanities. Through the construction of a dynamic relational database dedicated to French medieval hagiographic manuscripts (legendiers), based on the metadata of the Jonas database (IRHT), we demonstrate how modelling choices in the database are intertwined with philological Issues. Focusing on the composition, of legendiers, the article shows how standards-based modelling enables the analysis of textual variance, compilation logics, and manuscript relationships at multiple levels. By combining philological concepts with interoperable data structures, this work shows the importance of data modelling as an essential practice at the core of Digital Humanities research.
