A Diagnostic Model Using a Clustering Scheme

Abstract

It has been recognized that it is a challenging problem to deal with the situation where learners have diverse computing backgrounds and the learning content to be covered is also in the broad coverage. In the case, it’s required to devise a sophisticated diagnostic model for applying a proper teaching-learning method. We have drawn a scheme which can be applied to that case efficiently by using clustering algorithms based on web technology. In our approach, we focus on finding out methods for classifying both learners and learning content on the web. To make classification and manipulation of learning content ease, we reorganize learning content in order to have discrete form by introducing the concept of the knowledge unit which is extracted from each topic. Also, to make classification and diagnostic ease, we develop questions to measure them and analyze each question using item response theory (IRT) on the web. From the experiment of students sampled using our method, we show that learners with various backgrounds and the learning content with distribution on the broad range can be categorized effectively into the groups with homogeneous property. Also, we describe how to apply our proposed scheme to the introductory courses at postsecondary level.