Nghia Duong-Trung

Eye-tracking and AI for Enhanced Teaching (EYE-TEACH)

Funded by the HORIZON RIA Europe Program.
Duration: 01.2025 - 12.2027
Project Team:
- (Consortium lead) University of Turku, Finland
- Universitat De Valencia, Spain
- Universiteit Antwerpen, Belgium
- Open Universiteit Nederland, Netherlands
- Acrosslimits LTD, Malta
- Consiglio Nazionale Delle Ricerche, Italy
- Deutsches Forschungszentrum fur Kunstliche Intelligenz GmbH, Germany
- Centrum Nauki Kopernik, Poland
- Intralineas Education SL, Spain
- Blickshift GmbH, Germany
- Lexplore AB, Sweden

Project Summary

Recent education assessments (PIRLS, PISA) show a significant decline in reading comprehension amongst students in many European countries. The EYE-TEACH project is an initiative integrating artificial intelligence (AI) and eye-tracking (ET) that aims to transform educational practices and empower European teachers with new pedagogical skills for assessing and supporting their students' reading comprehension. The project addresses issues such as the post-COVID-19 educational challenges, mounting teacher workload and teacher shortages across Europe.

Eye tracking, although not without limitations, has proven its worth in educational research, offering insights into students' reading behaviours, cognitive load, and emotional engagement. It differentiates between various levels of reader comprehension, making it a valuable tool for personalised education. Complex eye-tracking data translated into intelligible output by AI and supported by a robust ethical and data privacy framework can provide teachers with actionable insights, aiding the selection of effective pedagogical strategies.

EYE-TEACH reaches beyond the state of the art by involving teachers and education practitioners early in the development process. By mapping educators' acceptance and readiness to adopt such novel technologies and highlighting the role of perceived ease of use and usefulness, the project has real potential to transform educational practices. The final joint result of the project will be a comprehensive resource of training materials and guidelines for using and implementing this technology in education, including benefit and risk assessments, and considerations of privacy and ethics issues.

By empowering teachers with an AI-assisted ET-analytics tool and training materials, the project aims to contribute positively to the future of education in Europe.

Publications (related to DFKI):

Linh Le, Quoc-Toan Nguyen, David Williams-King, Nghia Duong-Trung (2025). Time-Series Grid Encoding of Eye-Tracking Data for Explainable AI in Dyslexia Detection. ACM Symposium on Eye Tracking Research & Applications (ETRA 2025). May 26-29, 2025, Tokyo, Japan. Conference Rank B.

Towards Trustworthy AI. Pipeline Preparation, AI Explainability and Interpretability (AIX) Integration, Federated Learning Implementation, and Advanced Security Solutions

Funded by Saint Joseph University of Beirut, Lebanon.
Role and Participation:
- Project collaborator and PhD supervisor: 01.2025 - 09.2027

AI.EDU Research Lab 2.0: Support of Competencies for Research-based Learning Using Recommender Methods.

Funded by the CATALPA - Center of Advanced Technology for Assisted Learning and Predictive Analytics at the FernUniversität in Hagen.
Role and Participation:
- Senior Researcher: 01.2024 - present

The AI.EDU Research Lab is exploring the use of AI in university teaching. In version 2.0, the research focuses on supporting students' competencies, especially in deriving a term paper topic and a related guiding question with recommender systems (RecSys) and generative AI tools. For this purpose, the project is based on the results and experiences of the first research funding.

RecSys, based on different recommender methods, is used as a context-bound combination of AI technologies and didactic design for the purpose of transmitting recommendations to educational stakeholders. In the project, they are used to research and evaluate suitable AI methods to support students in finding a topic and generating a guiding research question for their term paper. A central research topic is, among other things, the transparency and trustworthiness of self-developed AI systems and those already in use. Comparatively, current tools and tasks for innovative applications with generative AI are explored.

This project is a collaboration with the Research Center CATALPA - Center of Advanced Technology for Assisted Learning and Predictive Analytics, FernUniversität in Hagen.

Software:

Term Paper Recommender System for FernUniversität in Hagen. Link. Under evaluation duration: 01.11.2024 - 31.12.2024.
Literature Recommender System for Term Papers. Link. Prototype.
KaggleGPT: Prompt-based Recommender System for Efficient Dataset Discovery. Link. Prototype.

Publications (related to DFKI):

Xia Wang, Niels Pinkwart, Nghia Duong-Trung, Claudia de Witt (2025). General LLM and Knowledge Powered Recommender System with a Case Study. 20th International Conference on Intelligent Systems and Knowledge Engineering (ISKE 2025). November 21-23, 2025, Shunde, China.
Lars van Rijn, Silke Wrede, Xia Wang, Nghia Duong-Trung, Angelin Mary Jose, Claudia de Witt, Niels Pinkwart (2025). Scaffolding Recommendations for Students' Essay Planning. Frontiers in Education. link. Q2 SCImago. ESCI Web of Science.
Nghia Duong-Trung, Xia Wang, Rahul R. Bhoyar, Angelin Mary Jose, Silke Elisabeth Wrede, Lars van Rijn, Theresa Panse, Claudia de Witt, Niels Pinkwart (2025). Kompagnon für das Verfassen von Hausarbeiten. Die 23. Fachtagung Bildungstechnologien der Gesellschaft für Informatik e.V. (DELFI 2025). 08-11 September 2025. Freiberg, Germany.
Nghia Duong-Trung, Xia Wang, Rahul R. Bhoyar, Angelin Mary Jose, Silke Elisabeth Wrede, Lars van Rijn, Theresa Panse, Claudia de Witt, Niels Pinkwart (2025). Knowledge Scaffolding Recommendation System for Supervising Term Papers. 20th European Conference on Technology Enhanced Learning (ECTEL 2025). Springer LNCS. 15-19 September 2025. Newcastle and Durham, United Kingdom. Acceptance ratio 27%. Conference Rank B.
Nghia Duong-Trung, Xia Wang, Rahul R. Bhoyar, Angelin Mary Jose, Silke Elisabeth Wrede, Lars van Rijn, Theresa Panse, Claudia de Witt, Niels Pinkwart (2025). TPRS: AI-Assisted Research Topic Refinement for Distance Learners. 26th International Conference on Artificial Intelligence in Education (AIED 2025). July 22-26, 2025, Palermo, Italy. Conference Rank A.
Xia Wang, Nghia Duong-Trung, Rahul R. Bhoyar, Angelin Mary Jose (2025). LLM-based Literature Recommender System in Higher Education — A Case Study of Supervising Students’ Term Papers. 2nd International Workshop on Generative AI and Learning Analytics (GenAI-LA): Evidence of Impacts on Human Learning, 15th International Learning Analytics and Knowledge Conference (LAK'25). March 3-7, 2025, Dublin, Ireland. Conference Rank A.
Rahul Rajkumar Bhoyar, Xia Wang, Nghia Duong-Trung (2024). KaggleGPT: Prompt-based Recommender System for Efficient Dataset Discovery. EduRS - Recommender Systems in Education at DELFI Conference on Educational Technologies. September 09-11, 2024. Fulda, Germany.
Nghia Duong-Trung, Xia Wang, Milos Kravcik (2024). BloomLLM: Large Language Models Based Question Generation Combining Supervised Fine-tuning and Bloom's Taxonomy. In Proceedings of the 19th European Conference on Technology Enhanced Learning (ECTEL 2024). Springer LNCS. September 16-20, 2024. Krems, Austria. Conference Rank B.

Completed Projects

MILKI-PSY: Multimodal Immersives Lernen mit künstlicher Intelligenz für Psychomotorische Fähigkeiten (Multimodal Immersive Learning with Artificial Intelligence for Psychomotor Skills).

Funded by the Federal Ministry of Education and Research of Germany (BMBF). Grant number 16DHB4015.
Role and Participation:
- Senior Researcher: 09.2022 - 12.2024
- Project Lead: 05.2023 - 12.2024

A research consortium around the Cologne Game Lab at TH Köln is developing a learning environment with artificial intelligence to support the independent learning of psychomotor skills. Project partners are the Institute for Product Development and Design Technology (IPK) at TH Köln, the German Research Center for Artificial Intelligence (DFKI GmbH), RWTH Aachen University, the Leibniz Institute for Human Development and Educational Information (DIPF), and the German Sport University Cologne (DSHS).

Anyone who wants to learn a sport trains in new psychomotor skills; until now, this has required role models such as on-site teachers who explain, demonstrate, and assess specific procedures. As part of the joint research project MILKI-PSY, a consortium led by the Cologne Game Lab at TH Köln, is developing a learning environment with artificial intelligence (AI) to support the training process.

The MILKI-PSY project designs an innovative environment for independent learning of psychomotor skills. For this purpose, the correct movement sequences are recorded by trainers using cameras and sensors. A virtual avatar generated from this recording will serve as a model for the learners. For example, it can be displayed on a large screen in an augmented or virtual reality environment. With the help of artificial intelligence and automated error detection, the learning progress is analyzed, and individual feedback is generated.

Publications (related to DFKI):

Mai Geisen, Abhishek Samanta, Khaleel Asyraaf Mat Sanusi, Nghia Duong-Trung, Milos Kravcik, Stefanie Klatt (2024). Visual and Tactile Information Provision in Golf: Conceptualization of a Training Application Incorporating Wearable Devices and Extended Reality. In Proceedings of the 22nd IEEE International Conference on IT in Higher Education and Training (IEEE ITHET 2024). November 6-8, 2024. Paris, France. link. DOI.
Aleksandra Nekhviadovich, Michal Slupczynski, Nghia Duong-Trung and Stefan Decker (2024). Analyzing Exercise Repetitions: YOLOv8-enhanced Dynamic Time Warping Approach on InfiniteRep Dataset. iWOAR - 9th International Workshop on Sensor-Based Activity Recognition and Artificial Intelligence. ACM Digital Library Proceedings. September 26-27, 2024, Potsdam, Germany. link. DOI.
Abhishek Samanta, Hitesh Kotte, Patrick Handwerk, Khaleel Asyraaf Mat Sanusi, Mai Geisen, Milos Kravcik, Nghia Duong-Trung (2024). IMPECT-POSE: A Complete Front-end and Back-end Architecture for Pose Tracking and Feedback. In Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization (ACM UMAP 2024). ACM Digital Library. 1 - 4 July 2024. Cagliari, Sardinia, Italy. link. DOI.
Hitesh Kotte, Florian Daiber, Milos Kravcik and Nghia Duong-Trung (2024). FitSight: Tracking and Feedback Engine for Personalized Fitness Training. In Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization (ACM UMAP 2024). ACM Digital Library. 1 - 4 July 2024. Cagliari, Sardinia, Italy. Acceptance ratio 22.6%. link. DOI.
Hitesh Kotte, Milos Kravcik, Nghia Duong-Trung (2023). Real-Time Posture Correction in Gym Exercises: A Computer Vision-Based Approach for Performance Analysis, Error Classification and Feedback. MILeS 2023 International Multimodal Immersive Learning Systems Workshop at Eighteenth European Conference on Technology Enhanced Learning (ECTEL2023), September 4-6, 2023, Aveiro, Portugal.
Jan Schneider, Khaleel Asyraaf Mat Sanusi, Bibeg Limbu, Nghia Duong-Trung (2023). Novices make more noise! But how can we listen to it? MILeS 2023 International Multimodal Immersive Learning Systems Workshop at Eighteenth European Conference on Technology Enhanced Learning (ECTEL2023), September 4-6, 2023, Aveiro, Portugal.
Nghia Duong-Trung, Hitesh Kotte, Milos Kravcik (2023). Augmented Intelligence in Tutoring Systems: A Case Study in Real-time Pose Tracking to Enhance the Self-Learning of Fitness Exercises. In Proceedings of the Eighteenth European Conference on Technology Enhanced Learning (ECTEL2023). Springer LNCS. September 6-8, 2023. Aveiro, Portugal.

Tech4compKI: Personalized competence development through scalable mentoring processes.

Funded by the Federal Ministry of Education and Research of Germany (BMBF), the second phase of the Tech4comp project. Grant number 16DHB2208.
Role: Senior Researcher.
Participation: 09.2022 - 12.2024

The joint project tech4comp develops and researches design concepts to make the proven quality of individual mentoring for acquiring competencies measurable. In addition, digital tools are being developed in mentoring learning and exam rooms that teachers can use to supervise many learners and automate mentoring processes through AI-based knowledge services.

Project partners are the University of Leipzig; Technical University Dresden; German Research Center for Artificial Intelligence (DFKI GmbH); Martin-Luther-University Halle-Wittenberg; Technical University of Chemnitz, University of Technology; Economy and Culture Leipzig; Free University of Berlin; RWTH Aachen.

Publications (related to DFKI):

Hassan Soliman, Hitesh Kotte, Miloš Kravčík, Norbert Pengel, Nghia Duong-Trung (2025). Retrieval-Augmented Chatbots for Scalable Educational Support in Higher Education. 2nd International Workshop on Generative AI and Learning Analytics (GenAI-LA): Evidence of Impacts on Human Learning, 15th International Learning Analytics and Knowledge Conference (LAK'25). March 3-7, 2025, Dublin, Ireland.
Matteo Orsoni, Alexander Pögelt, Nghia Duong-Trung, Mariagrazia Benassi, Milos Kravcik, Martin Grüttmüller (2023). Recommending Mathematical Tasks Based on Reinforcement Learning and Item Response Theory. 19th International Conference on Intelligent Tutoring Systems (ITS2023). June 5-8, 2023, Corfu, Greece. Best Full Paper Award.

KIWI-biolab: International future laboratory for AI-based bioprocess development

Funded by the Federal Ministry of Education and Research of Germany (BMBF). link. Grant number 01DD20002A.
Task Force 1 (96 PM): Active Learning for High Throughput Data and Operation
Task Force 1 Members: Prof. Dr. Dr. Lars Schmidt-Thieme (University of Hildesheim, Germany), Dr. Stefan Born (TU-Berlin, Germany), Dr. Nghia Duong-Trung (TU-Berlin, Germany), Prof. Dr. Andrey Ustyuzhanin (Faculty of Computer Science, HSE University, Russia), Randolf Scholz (University of Hildesheim, Germany).
Role: Machine Learning Scientist
Participation: 12.2020 - 08.2022.

The KIWI-biolab – Künstliche Intelligenz für Wissensbasierte Integrierte Biolabore – brings international top scientists in artificial intelligence, machine learning and bioprocess engineering together to explore the opportunities and challenges of automation in biotechnological processes.

Our Future Lab aims to use AI and ML technologies to automatize data analysis on microorganisms and enable computers to plan and optimize experiments independently. The long-term target is to advance towards automatization for a more efficient and sustainable production.

The KIWI-Biolab is a 3 years initiative funded with EUR 4.5 million by the Federal Ministry of Education and Research (BMBF), within the Federal Government’s Strategy for the Internationalization of Education, Science and Research, and the Federal Government’s High-Tech Strategy 2025 frameworks.

Publications (related to TUB):

Nghia Duong-Trung, Stefan Born, Jong Woo Kim, Marie-Therese Schermeyer, Katharina Paulick, Maxim Borisyak, Mariano Nicolas Cruz-Bournazou, Thorben Werner, Randolf Scholz, Lars Schmidt-Thieme, Peter Neubauer, Ernesto Martinez (2022). When Bioprocess Engineering Meets Machine Learning: A Survey from the Perspective of Automated Bioprocess Development. Biochemical Engineering Journal, Biochemical Engineering in the Era of Machine Learning and Automation Special Issue. SCIE Web of Science. IF 4.446. Q2 SCImago. link. DOI.
Kiran Madhusudhanan, Johannes Burchert, Lars Schmidt-Thieme, Nghia Duong-Trung, Stefan Born (2022). Yformer: U-Net Inspired Transformer Architecture for Far Horizon Time Series Forecasting. European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD 2022).

Deployment of a learning management system for Can Tho University of Technology (CTUT)

Funded by CTUT, Vietnam
Duration:
- Phase 1: 7 months (From 03/2018 to 09/2018)
- Phase 2: 10 months (From 09/2019 to 06/2020)
Role: Project manager
Tasks: Budget management, human resource management, business analysis, courses structure development
Project members: Ha Le Ngoc Dung (co-manager), Pham Yen Nhi, Nguyen Trung Kien

In this project, we aim to deploy a learning management system that can be integrated into CTUT’s teaching and studying activities, through a combination of formal, social, mobile and experiential learning. It facilitates the management, delivery, engagement, and measurement of all CTUT’s elearning courses. The system is going to support more than 100 teachers and 4000 students yearly.

System deployment: https://elearning.ctuet.edu.vn/

Engineering and information technology development and application in aquaculture and fisheries

Funded by Japan's ODA. A cooperation between Can Tho University (Vietnam) and Tokyo University of Marine Science and Technology and Kyushu University (Japan)
Duration: 36 months (From 10/2018 to 09/2021)
Role: Machine learning developer
Tasks: Developing machine learning models for shrimp classification, shrimp's disease prediction, food recommendation. Deploying images database collection, sensor systems. Deploying mobile application. Conducting research papers.
Key project members: Assoc. Prof. Chi-Ngon Nguyen, Assoc. Prof. Truong Quoc Phu, Prof. Akio Okayasu, Prof. Tsuyoshi Ikeya, Assoc. Prof. Daisuke Inazu.

The overall objectives are to study, develop and apply of engineering technology and IT for sustainable development of shrimp farming in the Mekong Delta. The specific objectives are (i) acquiring the pond environment parameters for shrimp farming management, (ii) eExpert System development for aquaculture extension on mobile communication networks, (iii) using of renewable energy in shrimp ponds, and (iv) building an Information system for Aquaculture and Fisheries management.

Publications:

Nghia Duong-Trung, Luyl-Da Quach, Chi-Ngon Nguyen (2020). Towards Classification of Shrimp Diseases Using Transferred Convolutional Neural Networks. Advances in Science, Technology and Engineering Systems Journal (ASTESJ).
Nghia Duong-Trung, Luyl-Da Quach, Chi-Ngon Nguyen (2019): Learning Deep Transferability for Several Agricultural Classification Problems. International Journal of Advanced Computer Science and Applications (IJACSA) 10(1).
Nghia Duong-Trung, Luyl-Da Quach, Minh-Hoang Nguyen, Chi-Ngon Nguyen (2019): A Combination of Transfer Learning and Deep Learning for Medicinal Plant Classification, in Proceedings of ACM International Conference on Intelligent Information Technology (ICIIT 2019). Da Nang, Vietnam.
Nghia Duong-Trung, Luyl-Da Quach, Minh-Hoang Nguyen, Chi-Ngon Nguyen (2019): Classification of Grain Discoloration via Transfer Learning and Convolutional Neural Networks, in Proceedings of ACM International Conference on Machine Learning and Soft Computing (ICMLSC 2019). Da Lat, Vietnam. link. DOI. Best Presentation Award.