Symposium Programme

All timings are in BST (Aberdeen, United Kingdom). This is programme version 1.0.

Monday, 1st of June, 2026

Time Event Details
09:00 09:30 Registration  
09:30 09:45 RetroEval 2026 Symposium Welcome  
09:45 10:30 Keynote Speaker Sina Zarrieß
10:30 10:45 Coffee Break  
10:45 12:15 Oral Session 1: Evaluating NLG Architectures  
10:45 11:15 Oral 1-1: A Comparative Evaluation of End-to-End and Pipeline Approaches for Summarisation Fahime Same, Saad Mahamood, Srinivas Ramesh Kamath
11:15 11:45 Oral 1-2: RAG as a collapsed NLG pipeline Adarsa Sivaprasad, Barkavi Sundararajan, David M. Howcroft
11:45 12:15 Oral 1-3: Decomposition Does Not Help: Evidence from Semantic Clustering in LLM-based Causal Graph Discovery Nikolay Babakov, Alberto Bugarín-Diz
12:15 13:30 Lunch  
13:30 15:00 Oral Session 2: Rethinking Evaluation  
13:30 14:00 Oral 2-1: Never Truly Out of Fashion: A Retrospective Look at Evaluation in NLG Patrícia Schmidtová, Saad Mahamood, Ondřej Dušek
14:00 14:30 Oral 2-1: Oral 2-2: NLG Evaluation: Past, Present, Future Ehud Reiter
14:30 15:00 Oral 2-1: Oral 2-3: Solving the Task but Not the Problem: A Customer Support Case Study on Why Extrinsic Evaluation Matters Daniel Braun
15:00 15:30 Get to the bus!  
15:30 16:30 Bus to the castle  
16:30 17:45 Dunottar Castle Excursion  
17:45 18:15 Return to the bus!  
18:15 19:00 Bus back to Aberdeen  
19:30 21:30 Celebratory Dinner (location to be announced soon)

Tuesday, 2nd of June, 2026

Time Event  
09:00 09:30 Registration  
09:30 10:15 Keynote Speaker Beatrice Alex
10:15 10:45 Coffee Break  
10:45 11:45 Food for Thought  
10:45 11:15 FT-1-1: Evaluation and Assessment as Complementary Frameworks  
11:15 11:45 FT-1-2: Ehud Reiter and the University of Santiago de Compostela: some notes and memories  
11:45 13:00 Lunch  
13:00 13:45 Keynote Speaker Albert Gatt
13:45 14:00 Get your tea and coffee for the poster session!  
14:00 15:30 Poster Presentations  
—– —– Poster 1-1: Towards Grounded Evaluation of Multimodal Machine Translation Systems Sami Ul Haq and Sheila Castilho
—– —– Poster 1-2: Checking for implicit assumptions in data-to-text generation Kristýna Onderková, Ondrej Dusek
—– —– Poster 1-3: The Arabic Bible as an Evaluation tool: The Case Study of the Khalili Arabic Dialect Jakub Zbrzezny, Ehud Reiter, Wei Zhao
—– —– Poster 1-4: The NL4XAI program: A retrospective Jose Maria Alonso-Moral
15:30 16:30 Panel Discussions  
15:30 16:00 P-1-1: BabyTalk Albert Gatt, Yaji Sirpada, Saad Mahamood
16:00 16:30 P-1-2: Commerical NLG Iain Davy, Yaji Sirpada
16:30 17:45 Closing surprises 🤫