A clustering algorithm based on minimum spanning tree with e-learning applications

Abstract

The rapid development of web-based learning applications has generated large amounts of learning resources. Faced with this situation, clustering is valuable to group modeling and intelligent tutoring. In traditional clustering algorithms, the initial centroid of each cluster is often assigned randomly. Sometimes it is very difficult to get an effective clustering result. In this paper, we propose a new clustering algorithm based on a minimum spanning tree, which includes the elimination and construction processes. In the elimination phase, the Euclidean distance is used to measure the density. Objects with low densities are considered as noise and eliminated. In the construction phase, a minimum spanning tree is constructed to choose the initial centroid based on the degree of freedom. Extensive evaluations using datasets with different properties validate the effectiveness of the proposed clustering algorithm. Furthermore, we study how to employ the clustering algorithms in three different e-learning applications.