A Framework for Building Intelligent Tutoring Systems

Abstract

Intelligent tutoring systems provide customized instruction or feedback to learners, without intervention from a human teacher. This feature causes that intelligent tutoring systems attract attention because they allow learning everywhere, every time and the cost of courses is cheaper than traditional in-class learning. In this work we propose a formal framework for building intelligent tutoring systems. The particular elements of those systems such as: learner profile, domain model, methods for determination and modification of a learning scenario and for computer adaptive tests are presented. Additionally, we describe an application of rough classification in e-learning systems. The conducted experiments and analysis demonstrate that the personalization has a significant influence on a learning process and the probability of passing all lessons from the learning scenario is greater if the opening learning scenario is selected using a worked-out methods than chosen in a random way. The obtained results proof the correctness of our assumptions and have significant implications for development of intelligent tutoring systems.