Industry Day Programme

Time Presentation Authors
08:50 Opening Teresa Lynn, Natalie Schluter
09:00 Tiny Aya: Bridging Scale and Multilingual Depth Beyza Ermis (Cohere)
09:30 PronunCise Phoneme-level Speech Analysis Youssef Ouarrak (PronunCise)
09:40 LoCar: Localization-Aware Evaluation of In-Vehicle Assistants through Fine-Grained Sociolinguistic control Kiwoong Park (KAIST), Seogyeong Jeong (KAIST), Ken Freidl (BMW)
09:50 ELDA at the crossroads: bridging language resources and responsible AI Gabriele Chignoli (ELDA)
10:00 A Comprehensive Full-Form Lexicon for Computational Processing of Arabic Dialects Jack Halpern (The CJK Dictionary Institute)
10:10 Deep Adaptive AI Translation (DAAIT): Manuel Herranz (Pangeanic), Marina Albert (Pangeanic)
10:20 VisualGPT: A Multimodal AI Tool for DeafDidactic Materials in Vocational Training Jörg Rutenberg (OpenMind Software)
10:30 Mozilla Data Collective: The data platform for human agency and fair value exchange Robert Pugh (Mozilla Data Collective)
10:40 Coffee break and Poster
11:00 Equitable Voice Technologies For Real World Classrooms Emma O'Neill (AI Labs Curriculum Associates)
11:30 From Hype to Metrics: Assessing Agentic LLMs for Complex Analysis Stacey Bailey (MITRE), Christy  Doran (MITRE), Keith Miller (MITRE)
12:00 FUSE: Record‑Agnostic Multimodal Semantic Embeddings for Industrial Data William Dyer (Terumo Blood & Cell Technologies)
12:30 Short Break
12:40 Antonio Zampolli Prize Talk
13:20 Lunch
14:45 Invited Local Talk: Nicolau Dols: Catalan in Majorca: main features, challenges, and research tools Nicolau Dols
15:15 Short Break
15:30 Is Agentic RAG worth it? Pietro Ferrazzi (Fondazione Bruno Kessler)
15:40 AskRegula: Evidence-Grounded Regulatory QA with Integrated Benchmarking and Compliance-Oriented Evaluation Tuba Gokhan (Mohamed bin Zayed University of Artificial Intelligence)
15:50 From Zero-Shot to Production: Building Robust NER for Automotive Service Analytics Adeel  Zafar  (Halmstad University)
16:00 SNOFlow: An Interface for Human-Led Medical Entity Linking to SNOMED CT with Agent Assistance Demian Gholipour (Quantexa)
16:10 Building Trust in LLM Products: A Production-Ready Evaluation Framework Erin Pacquetet (SCIAM), Christophe Séjourné (SCIAM)
16:20 Efficient Knowledge Base Population with Large Language Models for Industrial Applications: A Space Mission Case Study. José Manuel Gómez-Pérez (Expert AI)
16:30 FREC: A Model Context Protocol Infrastructure at the service of SMEs in the Catalan Digital Ecosystem Lluís Padró (Process Talks S.L./ Universitat Politècnica de Catalunya)
16:40 The need for problem-first research for European industry and the public sector. Toms Bergamis (Tilde)
16:50 Coffee break and Poster
17:20 Word Is Not Only Text: The Word Benchmark and the Limits of Agentic Document Manipulation Adrian de Wynter (Microsoft/ The University of York)
17:40 Panel session Moderators: Teresa Lynn, Natalie Schluter
18:30 OpenEuroLLM: safe, diverse and open LLMs à la EU Gema Ramírez-Sánchez (Prompsit)
18:40 Transperfect/Dataforce Alexandros Poulis (Transperfect/Dataforce)
18:50 The HiTZ Chair of Artificial Intelligence\nand Language Technology Aitor Soroa (Hitz)