Back to the previous page

Towards Personalized Learning Material Generation with Hybrid AI

Capture d'écran 2023-10-13 231046

Funding Agency: Leading House for Latin America (mandated by the Swiss State Secretariat for Education, Research and Innovation SERI)

Duration: 2022-2023

Partners: Universidad Privada Boliviana (UPB) 

The COVID-19 pandemic is impacting educational systems all around the world. As a result, more than one billion children are at risk of falling behind. Artificial Intelligence will play an essential role in sustaining an active and engaging learning experience breaking the current educational paradigms. However, current state-of-the-art methods train AI models to mimic and not to be factual. Moreover, these models do not support the personalization for the students' needs. This project proposes a novel concept for automatically creating engaging learning material from videos and textbooks. The learning material includes quizzes, memory cards, and a question & answer system. This project expects to support students' learning experience and facilitate the teachers' diagnosis process of learning trajectories. We propose using Hybrid AI, i.e., combining data-driven learning approaches with knowledge representation, to generate engaging yet personalized activities to consolidate the learning process. Advanced semantic student digital twin and learning trajectory representations will empower interoperability while guaranteeing data privacy. The project's approaches will provide individualized profiling and personalized material based on knowledge graphs, natural language understanding and generation beyond the state-of-the-art. This project's main result would be a holistic approach to automatically creating learning material from videos and textbooks. However, in the long run, we aim to build a virtual learning coach based on conversational agents capable of easing learning.

Demos

  • Study-Buddy Video demonstrator.


News

  • 02.2023: Trip to Cochabamba. D. Calvaresi traveled to Cochabamba to consolidate the project scientific activities with the Bolivian team, and also organized a seminar at Universidad Privada Boliviana, focused on the discussion of digital learning tools, with the participation of the faculty and students. Moreover, a trip within Bolivia to the city of La Paz was organized, in order to meet the faculty of UPB in that campus, including the local authorities. A workshop was organized with several speakers including D. Calvaresi, in order to boost other potential collaborations with the local professors and academic managers.

  • 03 2023: Trip to Sierre. D. Collarana, F. Martinez, C. Florida, and M. Arispe spent a week visiting the Intitut of Informatics at HES-SO Valais-Wallis. The trip was mainly used to boost the development of the demonstrator and finalize writing the first paper. Furthermore, two seminars were organized to showcase the works of the Bolivian team in presence of researchers of HES-SO in the TechnoPôle campus. Another seminar was organized with Dr. Collarana as main speaker focusing on the development of Generative AI methods. Furthermore, a visit to the School of Engineering HES-SO at Sion was organized, hosted by Prof. Pamela Delgado, in order to explore potential collaborations. 

  • 04.2023: Trip to Cochabamba. J-P Calbimonte visited UPB in Cochabamba, with the focus on the finalization of the project, and the writing of the second publication, and discussion of potential follow-ups. Moreover, a visit to the FESA Fundacion Educacional San Agustin was organized, in order to explore other collaborations including direct involvement of local high schools in this type of research initiatives.
  • 30.05.2023 Demo Presentation of Diego Collarana at ESWC 2023
  • 30.06.2023 Final report successfully submitted.

Publications

  • Study-Buddy: A Knowledge Graph-Powered Learning Companion for School Students. Fernanda Martinez, Diego Collarana, Davide Calvaresi, Martin Arispe, Carla Florida and Jean-Paul Calbimonte. Extended Semantic Web Conference ESWC. Heraklion, Greece. June 2023. (Proceedings to appear) 
  •  QALD-9-ES: A Spanish Dataset for Question Answering Systems. Javier Soruco Lopez, Diego Collarana, Martin Arispe. 19th International Conference on Semantic Systems SEMANTICS. Leipzig, Germany. September 2023. (Accepted, Proceedings to appear