Modeling and Implementing Intelligent Educational Environments Using An Interdisciplinary Approach

May 2002

Proceedings of the Fifteenth International Florida Artificial Intelligence Research Society Conference (FLAIRS 2002)

Volume

Proceedings of the Fifteenth International Florida Artificial Intelligence Research Society Conference (FLAIRS 2002)

All Papers

This paper presents the results that we have been achieving with our research involving educational environments based on Intelligent Tutoring Systems (ITS) architecture using a MAS (Multi-Agent System approach. A general guideline was idealized to be use as a reference for our research group to design and to implement intelligent educational software. Specially, interactive educational software modeled as a game. We believe that now we have the agent’s technology that makes possible to build interesting solving problems environments. However, it is very important to remember that educational process is closely connected with country culture. Educational environments that work very well in a specific context will not necessarily perform in the same way in a different place. It can be viewed as a restriction in many ways: as applicability for other educational reality, pedagogical paradigm, and so on. However, we argue that educational process has a kernel of necessities and aspects to be observed. When you intend to work considering learner as central person in the educational process, under the learn-to-learn paradigm. So, it causes an immediately reflection in the way we design and model an educational software. We discuss our ideas and guidelines used to model and to implement our systems under pedagogical viewpoint. We explore the possibilities of designing such systems using a MAS (Muti-agent) approach in order to explore the possibilities of this technique. In our proposal the domain is modeled with reactive agents and some of them we applied techniques of Machine Learning (Reinforcement Learning). The students and the set of pedagogical agents (task agents, assistants, and tutors) were modeled as cognitive agents using BDI architecture (belief, desire, and intention). Due to the space we are not describing the systems in details, but exemplifying how the guidelines were used.