A Detection Model for E-Learning Behavior Problems of Student Based on Text-Mining

Abstract

Finding a solution to reduce academic risk is an often-required task in online and hybrid courses. It is insufficient for teachers to rely on intuition and experience to identify the students with potential academic risk in e-learning. Therefore, in this study, we propose a detection model to automatically detect e-learning behavior problems of students. We used web course as pre-class learning task in information technology and curriculum integration course in the fall semester of 2018 at Central China Normal University, and recorded one semester of online discussions and learning behavior traces, involved 78 students in total. The experimental results indicated that the detection model can effectively identify poor time management, academic procrastination, low participation, and dishonest behavior. The model is useful to identify the student’s negative emotion which lasted a certain amount of time, but insufficient to identify the short-term emotional state which depended on high classification accuracy. The detection model and these results give the researchers and teachers a view of early alert for students at risk of academic failure, and how to improve student success in e-learning.