Funded Projects
EMOSCI - French-German Project (ANR / DFG)
(2026-2029)
Consortium: médialab - Sciences Po (Paris Institute of Political Studies), University of Montpellier (France), i3S (France), GESIS - Leibniz Institute for the Social Sciences (Germany), University of Trier (Germany)
Context & Goals: Scientific claims and informal discussions about science-related issues online have increased significantly, often becoming tools for manipulative narratives and disinformation. Emotions like fear and pride play a key role in spreading such claims, especially during crises like pandemics or natural disasters. Even well-intended science communication introduces distortions through (over-)simplification, decontextualization, inaccurate or missing references, complicating fact-checking and broader discourse analysis. While emotions frequently coincide with these distortions, their interplay remains under-investigated.
EMO-SCI researches the link between emotions and informal online scientific discourse by integrating emotion detection with the analysis of informal scientific claims. It pioneers research at their intersection, offering new annotated data, models and tools for analyzing emotions in online science claims, while also exploring these phenomena for the first time in the context of dialogues with Artificial Intelligence (AI) chatbots.
The project’s goals are oriented towards providing tools for better understanding and reducing science-related mis- and disinformation online.
More at: https://anr.fr/Project-ANR-25-CE23-206
AI for Democracy: Democratic Commons (BPI France)
(2025-2027)
Consortium: médialab - Sciences Po (Paris Institute of Political Studies), Sorbonne University, make.org
Context & Goals: The project aims to define, identify, measure, and mitigate democratic biases through an interdisciplinary approach. Methodologically, it builds on a strong theoretical foundation of key theories of democracy (e.g., participatory, deliberative, agonistic) to identify democratic principles (e.g., inclusiveness, plurality) and develops methods to evaluate democratic biases as deviations from these defined standards during interactions between citizens and LLMs.
Ultimately, the “AI for Democracy” projects aims to establish a robust and systematic framework for assessing and mitigating democratic biases in LLMs, ultimately contributing to the responsible development of AI systems that support democratic values and enhance participatory democracy platforms.
More at: https://about.make.org/democratic-commons/landing-page
AI4SCI - French-German Project (ANR / DFG)
(2021-2024)
Consortium: University of Montpellier (France), GESIS - Leibniz Institute for the Social Sciences (Germany)
Context & Goals: Scientific discourse, as seen on social media and in news articles, has been shown to compromise the accuracy of scientific findings. Research has argued that, on social media, boundaries between fact-based science, common sense, everyday experience or even ‘gut feeling’ are blurred (Neuberger et al., 2021). Complex scientific claims are uttered in the form of short, digestible, and often inaccurate snippets (Dunwoody, 2021). This phenomenon has led to online scientific debates being uninformed and has conduced to controversy and polarization of online debates (Bruggemann et al., 2020). Examples include social media discussions about global health pandemics or climate change.
In the AI4Sci project, we aim at the establishment of a novel interdisciplinary research subfield: Scientific online discourse processing. Through a unified definition, multiple ground-truth corpora, empirical task formalizations, and baseline methods and models, the project aims to lay the foundation for the study of scientific online discourse processing as a distinct, well-defined field of NLP/IR research, with direct interdisciplinary applications.
For a list of publications, references and contributions, see my Doctoral Dissertation
For more information about the project, see: https://ai4sci-project.org/