Student Model and Clustering Research on Personalized E-learning

Abstract

As the internet and data mining technologies are developing rapidly, how to provide various students with high-quality education services has become the hotspot in the internet environment. In order to promote the characteristics of online education and enhance the quality of personalized e-learning, in this paper, we propose a novel algorithm named MK-means by exploiting the cluster-wise weighing co-association matrix mechanism and improving the K-means algorithm based on the mean shift theory. The experimental results on the UCI’s Iris and Wine test sets demonstrate its effectiveness and superiority, finding that the total F-measure of MK-means achieves better performance than the Hierarchical Clustering, FCM, K-means, SOM, and X-means algorithms. Finally, the new algorithm combined with the student model explains the clustering results in detail from perspectives of cognitive model and knowledge map respectively and can extend to support the personalized e-learning in a wide range.