Understanding MOOC Reviews: Text Mining using Structural Topic Model

Abstract

Understanding the reasons for Massive Open Online Course (MOOC) learners’ complaints is essential for MOOC providers to facilitate service quality and promote learner satisfaction. The current research uses structural topic modeling to analyze 21,692 programming MOOC course reviews in Class Central, leading to enhanced inference on learner (dis)satisfaction. Four topics appear more commonly in negative reviews as compared to positive ones. Additionally, variations in learner complaints across MOOC course grades are explored, indicating that learners’ main complaints about high-graded MOOCs include problemsolving, practices, and programming textbooks, whereas learners of low-graded courses are frequently annoyed by grading and course quality problems. Our study contributes to the MOOC literature by facilitating a better understanding of MOOC learner (dis)satisfaction using rigorous statistical techniques. Although this study uses programming MOOCs as a case study, the analytical methodologies are independent and adapt to MOOC reviews of varied subjects.