The SEME (semantics and information extraction) team is interested in the problems of accessing the meaning contained in language productions, for the purposes of analysis, comprehension, modeling or production. We apply our research to the written modality, without restriction on the original medium (text produced in electronic format, or from a speech transcription, or from optical recognition) and work on productions in open or specialized domains such as the medical field. We use both linguistic and statistical or neural learning approaches. We are particularly interested in the latter type of approach, and in the environmental costs they generate in automatic language processing, both during production and during use.
Information extraction
Corpus and modeling
Sémantics, poly-lexical expressions
The team comprises 10 permanent members (CNRS researchers, lecturers at Université Paris-Saclay, ENSIIE, and Université Sorbonne Paris-Nord), 14 PhD students, and 3 post-docs or fixed-term contracts. We maintain links with industry (theses under CIFRE contracts, research projects) and regularly organize scientific events (TALN conference, scientific workshops, etc.).
Laura Ascone, Lucie Gianola, Julien Longhi, Laurène Renaut. La linguistique forensique pour l’analyse du discours : anticiper les risques, aider à la décision, répondre aux menaces. Colloque R2DIP, « Les notions de risques, société et sécurité dans les discours institutionnels et politiques », CY Cergy Paris Université, Dec 2017, Cergy, France. ⟨hal-05682141⟩
Théophile Lenoir, Ana Valdivia, Aurélie Bugeau, Anne-Laure Ligozat. Beyond the Energy Efficiency Directive. Observatory on the Environmental Footprint of AIArtificial Intelligence, 2026. ⟨hal-05680820⟩
Iskandar Boucharenc, Eve Sauvage, Thomas Gerald, Julien Tourille, Sabrina Campano, et al.. Using Syntax for the Semantic Representation of Sentences. SLiDE 1st Workshop on Structured Linguistic Data and Evaluation at the 2026 Language Resources and Evaluation Conference (LREC 2026), May 2026, Palma de majorque, Spain. pp.169–179, ⟨10.63317/4gtinxarm3dd⟩. ⟨hal-05669816⟩
Clément Morand, Aurélie Névéol, Anne-Laure Ligozat. The Rising Unsustainability of AIArtificial Intelligence Graphics Cards Production. LIMITS 2026: 12th Workshop on Computing within Limits, Jun 2026, Online, France. ⟨hal-05666542⟩
Kim Gerdes. The Grammar Does the Work: Functional vs. Lexical Dependency Length Minimization Across Universal Dependencies. UDW 2026 – Ninth Workshop on Universal Dependencies, May 2026, Palma De MaJorque, Spain. pp.163-173, ⟨10.63317/4akqrtsv7i65⟩. ⟨hal-05676925⟩
Iskandar Boucharenc, Sahar Ghannay, Christophe Servan, Laure Soulier, Sophie Rosset. Étude de l’adaptation des gros modèles de langues par retour visuel. Journée Visu, GdR IG-RV, Jun 2023, Orsay, France. ⟨hal-05670004⟩
Emmett Strickland, Marc Evrard, Valentina Fedchenko. Transfer Learning for Creole TTS: A Pilot Study on Whether Substrate Phonologies or Lexifier Vocabularies Matter More. Towards Inclusivity and Equality: Language Resources and Technologies for Under-Resourced and Endangered Languages, SIGUL 2026 Joint Workshop with ELE, EURALI, and DCLRL, May 2026, Palma De Majorque, Spain. ⟨10.63317/5d5qjmokuvmc⟩. ⟨hal-05617449⟩
Clémentine Bleuze, Bruno Guillaume, Aurélie Névéol, Karën Fort. Omniprésents et anthropomorphisés : analyse lexico-syntaxique des discours sur les LLM. TALN 2026 – 33e Conférence sur le Traitement Automatique des Langues Naturelles, Jun 2026, Nantes, France. ⟨hal-05670834⟩
Clémentine Bleuze, Karën Fort, Vincent P. Martin, Aurélie Névéol. Grands modèles de langue pour prédire la santé mentale : une revue exploratoire de la documentation des biais et de l’utilité clinique. TALN 2026 – 33e Conférence sur le Traitement Automatique des Langues Naturelles, Jun 2026, Nantes, France. ⟨hal-05670826⟩
Clément Morand, Aurélie Névéol, Rosy Tsopra, Anne-Isabelle Tropeano, Sophie de Chambine, et al.. Prospectively Evaluating the Environmental Impacts of Digital Health Applications : A Case Study and Recommendations. Journal of the American Medical Informatics Association, 2026, ⟨10.1093/jamia/ocag091⟩. ⟨hal-05628404⟩
Thomas Gerald, Sahar Ghannay, Julie Lascar, Paul Lerner, Anne Vilnat. Can Multimodal LLMs Generate Pedagogical Questions?. LREC 2026, May 2026, Palma, Spain. ⟨10.63317/4z4gj3h8jmc7⟩. ⟨hal-05658326⟩
Thierry Hamon. Description of the LISN system for extracting terms. DEfinition and Term Extraction CHallenge 2026 (DETECH 2026), Jun 2026, Zadar, Croatia. ⟨hal-05669893⟩
Marie Schmit, Melvin Selim Atay, Khalid Belhajjame, Ulysse Le Clanche, Emmanuel Coquery, et al.. ShareFAIR-KG, a centralised knowledge base of scientific workflows. JOBIM 2026 – Journées Ouvertes en Biologie, Informatique et Mathématiques, Jun 2026, Strasbourg, France. ⟨hal-05666980⟩
Louis Estève, Marie-Catherine de Marneffe, Nurit Melnik, Agata Savary, Olha Kanishcheva. A survey of diversity quantification in natural language processing: The why, what, where and how. 2026. ⟨hal-05661565⟩
Alexandre Genadot, Nicolas Guilliot, Philippe Boula de Mareüil. Introduction to the book “Cartographier les Langues de Nouvelle-Aquitaine: entre Grammaire et Société”. 2026. ⟨hal-05662837⟩
Agata Savary, Manon Scholivet, Carlos Ramisch, Takuya Nakamura, Eric Bilinski, et al.. PARSEME 2.0 Multilingual Corpus of Multiword Expressions. LREC 2026 – 15th biennial Language Resources and Evaluation Conference, ELRA Language Resources Association, May 2026, Palma De MaJorque, Spain. pp.4819-4834, ⟨10.63317/2iy5qf38yhay⟩. ⟨hal-05661505⟩
Julie Halbout, Annelies Braffort, Michèle Gouiffès, Diandra Fabre, Julie Lascar. Learning to Spot Signs from Named Entities. A study on French Sign Language. LREC2026 12th Workshop on the Representation and Processing of Sign Languages: Language in Motion, May 2026, Palma de Majorque, Spain. ⟨10.63317/26i8n4zuyzyx⟩. ⟨hal-05636077⟩
Damien Lacroux, Aurélie Bugeau, Anne-Laure Ligozat. The indirect rebound effects of AIArtificial Intelligence as undone science: philosophical reflection on two structural causes. Undone Computer Science, Mar 2026, Luxembourg, Luxembourg. ⟨hal-05624399⟩
Benedictus Kent Rachmat, Thomas Gerald, Zheng Zhang, Cyril Grouin. Les données de calibration comptent-elles vraiment pour LoRA?. EvalLLM2026 : Atelier sur l’évaluation des modèles génératifs (LLM), le RAG et challenges, Jul 2026, Nantes (France), France. ⟨hal-05633638⟩
Mathilde Aguiar, Pierre Zweigenbaum, Nona Naderi. Assessing the Difficulty of Inference Types in Natural Language Inference for Clinical Trials. The Fifteenth Language Resources and Evaluation Conference (LREC 2026), May 2026, Palma, France. pp.5290-5300, ⟨10.63317/359toazp33g8⟩. ⟨hal-05652719⟩
Jenny Copara, Nona Naderi, Gilles Falquet, Douglas Teodoro. MeSH Concept Relevance and Knowledge Evolution: A Data-Driven Perspective. 12th International Conference on Information Management and Big Data. Communications in Computer and Information Science, Oct 2025, Lima (Pérou), Peru. pp.280-299, ⟨10.1007/978-3-032-20322-9_20⟩. ⟨hal-05625658⟩
Clément Morand, Aina Rasoldier, Paul Gay. Not up to its critical perspective on digitalization: A Descriptive Analysis of How Sustainability is Approached in the ICT4S Conference. ICT4S, Jun 2026, Berne, France. ⟨hal-05615744⟩
Fanny Ducel, Lucie Digoin-Caparros, Ibrahim Al Kotob, Shayan Ahmed Shariff, Binesh Arakkal Remesh, et al.. Les benchmarks sont une source de biais des LLM : MMLU, CommonSenseQA et MGSM au microscope. TALN 2026 – 33e Conférence sur le Traitement Automatique des Langues Naturelles, Jun 2026, Nantes, France. ⟨hal-05618509⟩
Louis Estève, Christophe Servan, Thomas Lavergne, Agata Savary. A Diversity Diet for a Healthier Model: A Case Study of French ModernBERT. 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), Jul 2026, San Diego, United States. ⟨hal-05599374⟩
Virgile Barthet. Extraction d’information et classification de textes cliniques pour la prédiction du risque de décès. Intelligence artificielle [cs.AIArtificial Intelligence]. Université Paris-Saclay, 2026. Français. ⟨NNT : 2026UPASG019⟩. ⟨tel-05599487⟩
Luc Pommeret, Thomas Gerald, Christophe Servan, Sahar Ghannay, Patrick Paroubek, et al.. Étude des propositionneurs multilingues : formalisation, évaluation et interprétabilité. CORIA-TALN, ARIA; ATALA, Jun 2026, Nantes, France. ⟨hal-05597666⟩
Manon Scholivet, Agata Savary, Carlos Ramisch, Eric Bilinski, Takuya Nakamura, et al.. Edition 2.0 of the PARSEME shared task on multilingual identification and paraphrasing of multiword expressions. Proceedings of the 22nd Workshop on Multiword Expressions (MWE 2026), Mar 2026, Rabat, Morocco. pp.254-275, ⟨10.18653/v1/2026.mwe-1.33⟩. ⟨hal-05588684⟩
Jean-Luc Gauvain, Abdel Messaoudi, Holger Schwenk. Language Recognition Using Phone Lattices. International Conference on Speech and Language Processing, Oct 2004, Jeju, South Korea. pp.1283–1286. ⟨hal-01434492⟩
Luc Pommeret, Thomas Gerald, Sophie Rosset, Patrick Paroubek, Christophe Servan, et al.. Les propositions atomiques : un pont entre approches neuronales et symboliques. Journée interprétabilité, GDR TALTraitement Automatique des langues, Mar 2026, Jussieu, Paris, France. ⟨hal-05575718⟩
Luc Pommeret, Thomas Gerald, Patrick Paroubek, Sahar Ghannay, Christophe Servan, et al.. LLM-based Atomic Propositions Help Weak Extractors: Evaluation of a Propositioner for Triplet Extraction. KG-LLM@LREC – Knowledge Graphs and Large Language Models, ELRA, May 2026, Palma De Majorque, Spain. ⟨10.63317/3kna3utavhgb⟩. ⟨hal-05572941⟩
Jean-Luc Gauvain, Gilles Adda, Lori Lamel, Fabrice Lefèvre, Holger Schwenk. Transcription de la parole conversationnelle. Revue TALTraitement Automatique des langues : traitement automatique des langues, 2005, 45 (3). ⟨hal-01434260⟩
Jean-Luc Gauvain, Gilles Adda, Martine Adda-Decker, Alexandre Allauzen, Veronique Gendner, et al.. Where are we in transcribing French broadcast news?. Eurospeech, Sep 2005, Lisbonne, Portugal. pp.1665-1668, ⟨10.21437/Interspeech.2005-544⟩. ⟨hal-01434245⟩
Hélène Bonneau-Maynard, Alexandre Allauzen, Daniel Déchelotte, Holger Schwenk. Combining Morphosyntactic Enriched Representation with n-best Reranking in Statistical Translation. HLT/NACL workshop on Syntax and Structure in Statistical Translation, Apr 2007, Rochester, United States. pp.65-71. ⟨hal-01434104⟩
Nicolas Boizard, Hippolyte Gisserot-Boukhlef, Duarte M. Alves, André F T Martins, Ayoub Hammal, et al.. EuroBERT: Scaling Multilingual Encoders for European Languages. COLM 2025 – Second Conference on Language Modeling, Oct 2025, Montreal, Canada. pp.1-28. ⟨hal-05226285⟩
Pierre Lepagnol. Petits modèles génératifs en contexte industriel : Adaptation par prompting avec peu de données. Intelligence artificielle [cs.AIArtificial Intelligence]. Université Paris-Saclay, 2026. Français. ⟨NNT : 2026UPASG011⟩. ⟨tel-05572429⟩
Karin Dassas, Cyrille Bonamy, Bruno Bzeznik, Emmanuelle Frenoux, Gaël Guennebaud, et al.. Estimer l’impact carbone des activités numériques d’une unité de recherche. CNRS (EcoInfo). 2026. ⟨hal-05568070⟩
Ayoub Hammal, Pierre Zweigenbaum, Caio Corro. KAD: A Framework for Proxy-based Test-time Alignment with Knapsack Approximation Deferral. EACL 2026 – 19th Conference of the European Chapter of the Association for Computational Linguistics, Mar 2026, Rabat, Morocco. pp.3854-3872, ⟨10.18653/v1/2026.eacl-long.179⟩. ⟨hal-05571208⟩
Clément Morand, Jacques Combaz, Aurélie Névéol, Anne-Laure Ligozat. When rebound effect is not a side effect: analyzing sociotechnical contexts of digital technologies. 2026. ⟨hal-05566029⟩
Natalia Grabar, Cyril Grouin. Year 2021: COVID-19, Information Extraction and BERTization among the Hottest Topics in Medical Natural Language Processing. IMIA Yearbook of Medical Informatics, 2022, 31 (01), pp.254-260. ⟨10.1055/s-0042-1742547⟩. ⟨hal-03931852⟩
Pierre Lepagnol, Sahar Ghannay, Thomas Gerald, Christophe Servan, Sophie Rosset. Format Matters: A Critical Evaluation of Output Formats for Prompting LLMs in SLU and NER. The Fifteenth biennial Language Resources and Evaluation Conference (LREC 2026), May 2026, Palma de Majorque, Spain. ⟨10.63317/3osjjdr778fh⟩. ⟨hal-05546569⟩