Projects

Current projects

ANR Medialex Project (2021–2025, funded by ANR)
Partners: SciencesPo, CRESST (Polytechnqiue), INA
Abstract: How do transformations of the public and media spheres reshape the dynamics of emergence, imposition, and circulation of themes in public debate? Media transformations, particularly those linked to the rise of social networks, have enabled a wider range of actors to feed topics into public debate. The MEDIALEX project seeks to develop a computational approach to the dynamics of influence between legislative and media agendas.

European Pathfinder Astound Project (2022–2025, funded by the EC)
Partners: Universidad Politécnica de Madrid, INDEEP AI, Universitätsklinikum Hamburg
Abstract: The ASTOUND project (2022–2025), funded by the European Innovation Council, develops new approaches to conversational AI by integrating attention mechanisms and long-term memory, while placing a strong emphasis on ethical design, trust, and responsible use of socially interactive systems.

The GénéAlgoPsy Project (2023–, initially funded by a seed grant from BnF in 2023; self-funded since then)
Partners: Universté Paris-Cité (Astrid Chevance, UMR Thalim
Abstract: The GénéAlgoPsy project (Cultural Genealogy of Mental Pain) investigates the emergence and evolution of uses and representations of the notion of mental pain, a concept still poorly defined compared to physical pain. Led by psychiatrist and public health researcher Astrid Chevance at CRESS (INSERM/Université de Paris/APHP), it aims to clarify how mental pain is understood and differentiated from both the psychological dimension and psychological causes of physical pain.

Recent Past Projects

OuPoCo (Ouvroir de Poésie Combinatoire). (2020–2022)
Partners: Internal project
Abstract: This project aimed to produce sonnets by recombining lines from 19th-century sonnets, thus reviving an idea by Queneau (One Hundred Thousand Billion Poems). A “poetry box,” a low-energy autonomous poetry-producing machine, was created by the Raffard-Roussel studio as part of this project, along with a video produced by Riva Illustrations.

LAKME (Linguistically Annotated Corpora Using Machine Learning Techniques).
**Partners: EPHE, ENC (Funding by PSL)
Abstract: this project explored new techniques for morphosyntactic and syntactic analysis of text corpora in morphologically rich languages (such as Medieval French, Rabbinic Hebrew, and several Finno-Ugric languages (mostly Finnish and Fennic languages).