Le projet de consortium pluridisciplinaire HISINUM, soutenu par l’Académie 5 depuis 2023, a permis la fédération de trois équipes de recherche, quatre laboratoires et des partenaires institutionnels et internationaux dans une démarche structurante et ouverte afin de mutualiser des outils liés, interopérables et généralisables visant le renouvellement des pratiques de la recherche et de la question de la donnée en sciences humaines et sociales.
g PROGRAMME :
9h30 : Accueil et échanges sur les activités du consortium Hisinum
10h : Conférence. Andrea Loettgers, Research fellow at the University of Vienna : « Investigating the transdiciplinary application of model templates through projective methods »
Abstract
One of the most striking characteristics of contemporary modeling practice is the widespread use of the same basic equations, functional forms, algorithms, and quantitative methods across disciplines. Recent philosophical work on templates and template transfer have underscored the remarkable cross-disciplinary applicability of certain mathematical structures and computational algorithms. In this context, we introduced the concept of the model template, which captures the interplay between a mathematical structure, its associated computational tools, and the theoretical concepts that together represent a general mechanism. This mechanism can be applied to any domain or field that exhibits specific patterns of interaction.
An illustrative example of a model template is the scale-free network model. This model has been widely applied across disciplines, particularly in the analysis of large datasets, including their topology and properties such as their robustness to the removal of network nodes. However, tracing model templates through the vast body of recent scientific literature, in which they arguably play an increasingly constitutive role, is a challenging, if not impossible, task.
This has led us to develop digital methods, which allow philosophers and historians of science to interactively unfold the history of a model template and to build adequate intuitions about the patterns that govern their global adoption and distribution. The application of Large Language Model-aided approaches allows us to move beyond the analysis of selected local case studies, which, while undoubtedly of great importance to the history and philosophy of science, struggle to adequately treat contemporary modeling practices in interdisciplinary contexts.
