Evaluation for e-Learning Website of Physics by Browsing Path Analysis and Cluster Analysis of Access Log

Abstract

KIT Physics Navigation, a self-adaptive e-learning website of physics covering study contents for high school and university students, was published on the web in March 2016. It was built on the concept that “one web page should contain one topic”. For the first time, the access log analysis was performed on this website by examining how visitors browsed the webpages and deepened their understandings. It is noted that this analysis was carried out by using only the access logs acquired from the visitors who had browsed a webpage entitled “Uniformly accelerated linear motion” at least one time to extract the browsing path of the visitors who had an interest in the topic of the webpage. As a result, it was found that most of the visitors deepened their understandings of physics in stages by browsing from the web pages about fundamental topics to those about advanced topics. Furthermore, cluster analysis, which is widely known as the unsupervised learning method of machine learning, was performed on this website. Here, Ward’s Method was applied, and the variables were the number of visits and the visit duration. The result showed that the web pages about the following topics, “Derivation of uniformly accelerated linear motion from graph” and “Derivation of uniformly accelerated linear motion by using integration”, was classified as the group which had a large number of visits and long visit duration by the dendrogram. In the future, the websites need further improvements based on the results of these analyses.