SCOPE

Digital twin technology integrates the physical and digital worlds to accurately represent real-world systems, predict future conditions, and support optimal decision-making. This workshop promotes collaboration among researchers in artificial intelligence, mathematical modeling, and data assimilation to advance predictive analytics and intelligent decision support. The program focuses on three core themes: Pattern Recognition and Prediction, Physics-Based Modeling and Numerical Analysis, and Data Assimilation and Multimodal Data Integration. By bringing together scientists and practitioners from diverse disciplines, the workshop aims to deepen understanding of the fundamental technologies underpinning next-generation digital twins. It also provides a forum for exchanging ideas, addressing current challenges, and exploring future research directions, thereby fostering innovation and progress in this rapidly evolving field and its applications.

TOPICS OF INTEREST

KEYNOTE SPEAKER

User Models as Explainable Digital Twins for Human-Centred AI

Maria Virvou University of Piraeus, Greece Greece

Short Bio
She is Co-Founder and Co-Editor-in-Chief of the Springer book series Learning and Analytics in Intelligent Systems and Artificial Intelligence–Enhanced Software and Systems Engineering, as well as Co-Founder of the IEEE Information Intelligent Systems and Applications (IISA). She previously served as Editor-in-Chief of SpringerPlus and currently serves as Associate Editor of Knowledge-Based Systems (Elsevier). Professor Virvou holds a Ph.D. in Artificial Intelligence from the University of Sussex, funded by a competitive scholarship in Artificial Intelligence from the Greek State Scholarships Foundation (IKY) following national examinations. She also holds an M.Sc. in Computer Science from University College London (UCL) and a Degree in Mathematics from the National and Kapodistrian University of Athens. Her scholarly record includes over 400 publications and 8 monographs indexed in Scopus. She ranks 1st worldwide in Intelligent Software and Educational Software and 2nd in User Modeling, according to Scopus and Microsoft Academic Search. Professor Virvou has supervised 15 Ph.D. theses and 10 postdoctoral researchers and is currently supervising 5 Ph.D. candidates. She has coordinated numerous national and international research projects and has received multiple distinctions, including a UNESCO award for her contributions to computer science. In recognition of her impact, Springer published Advances in Intelligent Healthcare Delivery and Management: Research Papers in Honour of Professor Maria Virvou. She is consistently listed among the top 2% of researchers worldwide in Artificial Intelligence, according to the global academic ranking by Stanford University.

Abstract
As Artificial Intelligence systems evolve into collaborative partners, Human-Centred AI requires systems that can adapt to people while supporting mutual understanding, calibrated trust, and meaningful user control. User models provide a foundation for personalisation, but their potential extends to representing how a person’s interaction with AI changes over time. This keynote proposes advancing user models as Explainable Digital Twins (xDTs): evolving digital representations of a human collaborator, updated through interaction evidence and user feedback. These representations connect past interactions, current states, and possible future scenarios to support adaptive human–AI collaboration. The approach draws on VIRTSI—Variability and Impact of Reciprocal Trust States towards Intelligent Systems—a framework we developed for modelling human–AI trust dynamics. Using Deterministic Finite State Automata (DFAs), VIRTSI represents transitions between trust states, including justified trust, overtrust, and distrust. These transitions provide a structured basis for examining how users may interpret an interaction and how that interpretation shapes their subsequent responses. The keynote outlines a roadmap for connecting interaction evidence to trust-state updates, explaining the assumptions behind those updates, and enabling users to inspect and correct the model. It further explores how the proposed xDT could examine possible future trust trajectories under alternative AI responses—for example, acknowledging an error, explaining uncertainty, or requesting user input—and use these scenarios to guide adaptation. Central to this vision is bidirectional explainability: the AI explains its understanding of the user, and the user can challenge and refine that understanding. The aim is an evolving, explicitly partial representation that supports collaboration while preserving privacy and human agency.

IMPORTANT DATES

WORKSHOP PROGRAM COMMITTEE

Available soon.

PAPER SUBMISSION

PUBLICATIONS

AUTHOR'S KIT

SECRETARIAT CONTACTS