Guiando tareas mal definidas en ambientes virtuales de aprendizaje en 3d a través de tutores inteligentes.

Abstract

This doctoral thesis represents a significant advancement in the development of intelligent tutoring systems, focusing on the resolution of ill-defined problems in 3D Virtual Learning Environments. The research centers on the evolution of the MAEVIF architecture, resulting in MAEVIF-PID, a framework that incorporates an intelligent tutor designed to guide students in solving complex and poorly defined problems, using structured cabling as a case study. Exhaustive Systematic Literature Reviews (SLRs) were conducted to identify research gaps in 2D and 3D virtual learning environments and to understand the challenges associated with ill-defined problems. These reviews were crucial in developing a clear understanding of areas that require new solutions. The iPlus methodology was adapted and applied for the participatory development of learning content, which effectively represented illdefined problems through serious games. Additionally, the MASINA methodology was identified and applied to design a Multi-Agent Intelligent Tutoring System, achieving a practical approach to solving these problems in virtual environments. A key milestone of this research was the integration of artificial intelligence algorithms to generate adaptive solutions for ill-defined problems, highlighting AI's capability to provide relevant and effective responses. Finally, the rigorous validation of the applied methodological framework demonstrated its efficacy and robustness, confirming its utility in complex and dynamic educational contexts.