Séminaires du CENTAL

👤  Speaker :  ​ Ève Sauvage 

💬  Titre ​: Représentations des textes, entre approches neuronales et symboliques

Abstract : Je parlerai de comment transformer les entrées des modèles neuronaux, pour améliorer leurs performances ou pour diminuer la taille des entrées. Et, au-delà des modèles neuronaux, est ce qu'il est possible d'utiliser des approches symboliques pour obtenir ces mêmes conséquences ? Est-il possible de combiner les deux approches, quels sont leurs limitations et leurs intérêts ? l'IA neuro-symbolique permettra-t-elle de mieux représenter les textes ?

📅  Date  :  20/02/2026

📍 Local  :  Maison Des Langues A118 (FIAL)  ( Google Map link ) 

👤  Speaker :  Matthew Shardlow

💬  Titre ​:  Operations based Text Simplification

Abstract : In this talk, we will look at recent research on text simplification using techniques from Natural Language Processing and Generative AI. We will consider the types of linguistic operations that can be performed with a text to make it easier to read for an end user. We will also consider how current trends in research on text simplification are progressing to enable better evaluation and automated methodologies for making texts easier to read by considering the operations necessary to simplify them.  The talk will cover Dr. Shardlow’s research on text simplification over the past 10 years and will also draw on recent related research from the NLP domain.

Bio : Dr. Matthew Shardlow is a Reader in Natural Language Processing in the Department of Computing and Mathematics at the Manchester Metropolitan University. He was previously a member of the National Centre for Text Mining working on a Horizon 2020 funded project and studied his PhD at the University of Manchester under an EPSRC funded centre for doctoral training. He currently leads projects with industry partners including international publicly traded companies, charities and local government. He is an organiser of the Text Simplification, Accessibility and Readability workshop (EMNLP 2022, RANLP 2023, EMNLP 2024, EMNLP 2025), the SemEval-2021 shared task on Lexical Complexity Prediction (ACL 2021), The TSAR-2022 shared task on lexical simplification (EMNLP2022) and the BEA-2024 MLSP shared task (NAACL 2024). His research interests lie in the field of natural language processing and more recently generative AI. He has previously worked on topics including named entity recognition,  event extraction, machine translation, emoji semantics, text generation and has more recently explored phenomena of consciousness and anthropomorphisation in association to LLMs.

📅  Date  :  06/03/2026

📍 Local  :  UCLouvain Faculty of Law and Criminology (More 70) ( Google Map link ) 

👤  Speaker :  Alexandra Courtaux 

💬  Titre ​: Linguistes dans le monde du travail

Abstract : Dans la banque d’investissement, une grande partie de l’information pertinente provient de documents textuels tels que les rapports financiers, les contrats de prêts et l’actualité des marchés économiques. Ce cours présente le rôle du traitement automatique des langues et des modèles d’intelligence artificielle dans l’analyse et l’exploitation de ces données. Il abordera la place croissante de ces technologies dans le secteur, les principales approches techniques utilisées, ainsi que quelques cas d’usage concrets. Le cours évoquera également certains défis liés à leur utilisation, notamment d’ordre réglementaire, avec l’arrivée de l’AI Act, et énergétique, en raison des ressources nécessaires au développement et à l’utilisation de ces modèles.

📅  Date  :  20/03/2026

📍 Local  :  Auditoires UCLouvain Agora (AGOR 01) ( Google Map link )

👤  Speaker :  ​Rodrigo Wilkens

💬  Titre ​: Multiword Compositionality as a Lens on Language Models

Abstract : Multiword expressions (MWEs) are a source of difficulty in Natural Language Processing, as their meaning often depends on convention rather than simple composition. Expressions such as cheese knife illustrate that combining words does not necessarily yield additive meaning, while others, such as loan shark or kick the bucket, demonstrate varying degrees of semantic opacity. This talk examines how MWE meaning can be characterised, focusing on the contribution of individual components and the variability observed across expressions. Compositionality is treated as a continuum, reflecting asymmetric contributions among MWE components and the role of conventionalised usage. The discussion also considers how computational models handle MWEs in comparison with more regular linguistic patterns, highlighting where their behaviour remains sensitive to non-compositionality and variation. Recent reasoning-based models are considered in this context, with particular attention to whether more explicit reasoning is associated with improved handling of multiword meaning. The perspective is then extended beyond text, considering multimodal settings where meaning depends on grounding in visual context. Overall, the talk invites reflection on the role of MWEs as a controlled setting for examining how computational models process meaning.

 📅  Date  :  10/04/2026

📍 Local  :  Maison Des Langues A118 (FIAL)  ( Google Map link ) 

👤  Speaker :  ​Joakim Nivre

💬  Titre ​: Perspectives on Universal Dependencies

Abstract : Universal Dependencies (UD) is a project developing cross-linguistically consistent morphosyntactic annotation for many languages, with the goal of supporting multilingual research in natural language processing and linguistics. Since UD was launched in 2014, it has grown into a large community effort involving over 700 researchers around the world, together producing treebanks for over 180 languages. In this talk, I will first give a brief overview of the UD framework and its resources, and then present two ongoing projects related to UD. The first is MultiBLiMP, a massively multilingual benchmark for linguistic evaluation of language models, constructed by leveraging resources from Universal Dependencies and UniMorph. The second is the UD Constructicon project, which aims to develop a construction-oriented version of the UD guidelines and at the same time evaluate the typological validity of the UD framework.

📅  Date  :  08/05/2026

📍 Local  :  Maison Des Langues A118 (FIAL)  ( Google Map link )