Comparing Short-Term and Long-Term Online Courses Using the Kano Model and Neural Network Language Models

Abstract

Evaluating rich data from online courses poses challenges, even with established methodologies. A mixed methods approach can help in addressing research questions. One of the difficulties is to understand users' expectations before taking a course and their consumption experience after completing a course. The focus of this study is on the difference in short-term courses, which can be taken within a day, and long-term courses, which span several weeks or months. This paper compares two online courses for undergraduate students in Japan. The courses were hosted on the same system, one that was complete within 90 minutes (introduction to photography) and one that was conducted for a semester totaling 15 weeks (introduction to programming). Students were surveyed on the same 12 features related to online course satisfaction before and after each course. Textual comments were also gathered. The Kano model from customer satisfaction research was used to perform an ex-ante and ex-post comparative analysis for the 12 features of both short-term and long-term courses. A simple neural network was trained on freeform comments for both courses to create language models (Word2Vec) and compare the findings with the Kano model results.