معرفی
Djamé Seddah is a senior researcher at INRIA Paris (Almanach team), previously holding a tenured associate professor position at Sorbonne University. His research focuses on natural language processing (NLP), particularly multilingual syntactic analysis, syntax-semantics interfaces, and language models for low-resource languages. He has developed influential models like CamemBERT and contributed to annotated corpora such as the Sequoia corpus and French Social Media Bank. Current work emphasizes ethical AI, including detecting harmful content and mitigating language model biases.
Education: PhD in Natural Language Processing from Loria (Nancy, France), postdoc at DCU's GramLab (Dublin). Key achievements include defending an HDR in 2024 and leading projects like the BPI "Commun Numériques" initiative. Awards include the PSAI Prairie 2 Chair. Active in organizing conferences like SyntaxFest 2019 and shared tasks at IWPT.
Research spans parsing user-generated content, sociologically aware language models (SALM), and collaborative work with institutions like LightOn and Software Heritage. His lab's work addresses challenges in dialectal languages, noisy data, and ethical AI applications.
