Semantic technologies in e-learning: learning analytics and artificial neural networks in personalised learning systems

Abstract

The paper presents semantic clustering and artificial neural network (ANN) based learning analytics software agent for personalised adaptive multi-agent learning system. First of all, systematic literature review on application of ANN to personalise learning in Web of Science database was performed. After that, methodology of application of ANN and semantic clustering of learning material in personalised adaptive multi-agent learning system is presented. In the paper, personalisation in multi-agent learning system is based on learning styles model that requires the use of psychological questionnaire to determine student’s learning styles. The results of filling in the questionnaire could be incorrect since some students may answer the questionnaire dishonestly, irresponsibly, or make mistakes in self-diagnosis. This results in creation of an incorrect student’s model. This causes the system to provide suboptimal learning objects and scenarios to the student. The authors present a model of ANN based learning analytics agent to be used in the system. Proposed agent uses ANN to associate students’ learning styles with their real behaviour within the learning environment. The key factor describing the behaviour of students within the system is the learning content they seek out independently. Clustering the visited documents based on semantic content categorizes students into groups. Belonging to a semantic cluster is one of the inputs that can be used to train ANN agent. After training, the agent could identify potentially faulty student models by looking for anomalous behaviour for those learning styles. Such problems can be resolved by providing alternative learning objects or scenarios to the students and observing their choices. 1