Intelligent Learning Model Based On Significant Weight Of Domain Knowledge Concept For Adaptive E-Learning

Abstract

In order to support personalized learning, an adaptive learning system should have a capability to provide each student with a suitable learning material regarding his profile. However, the issue of student varieties in acquiring every Domain Knowledge Concept (DKC), and a range of DKC important variations in a particular learning material produced a complex dependency that causes a difficulty in the learning material selection process.  Existing rule-based learning material selection approach requires the definition of a huge set adaptation rules. However, this approach usually results in inaccurate and incorrect selection due to the inconsistent, insufficient and confluence of the defined rules.  Consequently, the process of learning material selection is hard to be algorithmized, therefore, intelligent methods are applied to handle the complexity challenges. This research proposes a significance weight approach that represents the complex dependency of learning material selection problem to substitute the rules definition in the selection process. In addition, this research proposes an intelligent learning model that combines unsupervised and supervised machine learning techniques to accurately select the learning material for a particular student adaptively. The unsupervised machine learning technique is vital in obtaining a learning material classification and labelling based on the proposed significance weight. Meanwhile, the supervised machine learning technique, the Multilayer Perceptron Artificial Neural Networks conducts the adaptation process that will assign the student to suitable learning materials regarding his performance upon specific DKC. With 98% achievement of classification accuracies, this model can be considered as highly accurate in selecting a correct and suitable learning material based on student’s domain knowledge level.