PhD position in NLP (M/F) – Evaluating and Controlling Automatic Easy-to-Read Simplification: Emotional Fidelity and Discourse Coherence (ÉCOFALC)

Descrizione

PhD position in NLP (M/F) – Evaluating and Controlling Automatic Easy-to-Read Simplification: Emotional Fidelity and Discourse Coherence (ÉCOFALC) Sintesi in italiano (traduzione automatica): L'organizzazione offre una posizione di dottorato in Linguistica Computazionale, focalizzata sulla valutazione e il controllo della semplificazione automatica dei testi per garantire la coerenza discorsiva e la fedeltà emotiva. Il progetto è interdisciplinare e co-supervisionato dai laboratori GREYC e MODYCO, con il supporto del laboratorio NIMH. Le principali mansioni includono lo sviluppo di modelli generativi per semplificare testi in un linguaggio facile da leggere, mantenendo l'engagement affettivo. È richiesta una laurea in Linguistica, Informatica o Psicologia Cognitiva. Il lavoro si basa su dataset e modelli già sviluppati e mira a creare strumenti per migliorare l'accessibilità testuale per pubblici neuro-atipici. The project is interdisciplinary, co-supervised by the GREYC and MODYCO laboratories, with support from the NIMH laboratory. It combines Natural Language Processing, linguistics, and cognitive psychology in order to: (i) enrich simplification corpora with discourse and emotional annotations; (ii) develop metrics for emotional fidelity and cognitive effectiveness; and (iii) validate the effects of the generated texts through behavioural and neurophysiological measures. The expected outcomes are new tools and resources for scientifically grounded textual accessibility, tailored to neuro-atypical audiences. This PhD project in Natural Language Processing aims to design generative models capable of automatically simplifying texts into Easy-to-Read and Understand language while preserving their discourse coherence and emotional content. The central hypothesis is that an accessible text should not only be simpler, but should also maintain affective engagement, which is crucial for attention, comprehension, and memorization. The work will build on the datasets and models developed at GREYC and disseminated in the following two publications: * Ledoyen F., Dias G., Maurel F., Chahir Y. Inclusive Easy-to-Read Text Generation for Individuals with Cognitive Impairment. In the 34th International Joint Conference on Artificial Intelligence (ECAI 2025). * Ledoyen F., Dias G., Pantin J., Maurel F., Lechervy A., Chahir Y. Facilitating Cognitive Accessibility with LLMs: A Multi-Task Approach to Easy-to-Read Text Generation. In The 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP 2025). Annuncio in inglese. Fonte: Euraxess (Commissione europea).