Preparing the past for the future: Curating a daylight simulation model of Hagia Sophia for modern data infrastructures

Abstract

Digital humanities and artificial intelligence applications rely on structured sets of data and metadata. Larger and more complex dataset as a product of growing com puting power and widely available advanced tooling demand modern data-management within projects, while scientific transparency, reusability, and interoperability demand machine-readable publishing and linking of project data. The latter implies infrastructure for long-term storage and AI-assisted search. Emerging platforms as the National Re search Data Infrastructure (NFDI) and Specialised Information Services (FID), both sponsored by the German government provide suitable repositories but require the cu ration of research data. This paper examines the interdependency of in-project data management and publishing and localizes possible AI-applications in in the context of a non-ideal case-study – the heterogeneous dataset of a light-simulation model of Hagia Sophia. It proposes a separation of the project-data into five scopes – raw data collec tions, reconstruction and material models, simulation environment, simulation results and digital publications – that allow to develop transferable solutions and integration into emerging infrastructures for reuse, search, linking and publishing of data between pro jects. The paper concludes that AI-applications in this context provide more general, transferable solutions for search and spatial image organisation within the research in frastructure and very specific solutions within a project. A standardised organisation of research data and metadata has to fit these applications.