Analyzing Learning Patterns Based on Log Data from Digital Textbooks

Introduction

With the development of e-publishing technologies and standards, it is easy to obtain digital books, such as “living books,” “talking books,” and “CD-ROM books,” from the Internet (Yin et al., 2015). Digital books have become a potentially effective pedagogic tool (Hezroni, 2004; Reinking, 1997; Snyder, 2002), indicated by the fact that digital book reading has increased significantly in the United States (Lee et al., 2012). Consequently, traditional textbooks are being increasingly replaced by digital textbooks (Ren, Uosaki, Kumamoto, Liu, & Yin, 2017).

Many researchers have been paying attention to the development of digital books to support teaching, learning, and scholarship. By using digital books, large bodies of log data can be accumulated, the analysis of which can be used to perform learning analytics.

Learning analytics can be used in making suggestions to policy makers, instructors, and learners (Baker & Inventado, 2014; Hwang, Hsu, Lai, & Hsueh, 2017). Therefore, learning analytics have become an important issue in education (Hwang, Chu, & Yin, 2017) that have entailed important changes in educational research.

The objective of learning analytics is to provide helpful information to optimize or improve learning designs, learning outcomes, and learning environments based on the analysis results (Greller & Drachsler, 2012; Hwang, Chu, & Yin, 2017).

In the analysis of learning behaviors in this study, we used a digital textbook system to collect students’ learning logs. Learning log is defined as a digital record of what learners have learned in a formal and an informal setting (Ogata, Hou, Uosaki, Mouri, & Liu., 2014; Mouri, Ogata, & Uosaki, 2015). The system was used in a commercial law course for undergraduate students, which was conducted in entirety in English: The students were assigned readings in English, and the teacher spoke in English. We also used questionnaires to collect data on students’ learning styles and cognitive styles.

Using these data, we applied the association analysis method with the Apriori algorithm to analyze students’ learning patterns or rules. One of the advantages of analyzing the learning patterns or rules is the preemptive prediction of students’ final grade and progresses in the future. As a result, teachers can improve their teaching strategies and support students’ learning behaviors. From the analysis, this study found three meaningful student types by considering the detected association rules.