A model for adaptive selection of learning material in an intelligent learning system using combination of supervised and unsupervised machine learning techniques

Abstract

One of the issues in personalizing student’s learning experience is the complexity to select the learning material that is suitable to the profile of each student such as his performance, level of knowledge, learning style and etc. A huge set of adaptatio n rules is required in order to design a highly adaptive learning system that can handle such complicated task. However, the adaptation rules definition task really depends on the effort of domain expert because only the domain expert is reliable in formul ating the scheme for adapting student’s profile and the learning material. We propose d an approach for intelligent and adaptive selection of learning materials that provide a consistent selection of suitable learning materials as well as reducing the depen dency on domain experts’ effort in adaptation rules definition, on the other hand, the domain expert point of view can still be encapsulated for a reliable domain knowledge concept representation. The proposed model treats the adaptation process as a super vised classification task that will assign student to suitable learning materials regarding to his performance upon specific domain knowledge concepts. Some experiments using K - means and Self - Organizing Map (SOM) for clustering and Artificial Neural Networ ks (ANN) for classification are implemented towards the learn ing materials data and student performance data as well. The experimental results have favo u rably shown that the proposed model using domain concept based clustering and classification to be prac tical and effective in solving the adaptive selection of learning material problem.