What can we learn from MOOC: A sentiment analysis, n-gram, and topic modeling approach

Abstract

The rapid change in the demands of the industry has become a phenomenon that has a widespread impact on the higher education landscape. Numerous research has been done to explain what institutions can do to overcome this phenomena. Generally, an institution’s internal efforts and government regulations are seen as solutions from past literature. However, the influence of public opinion on one’s choices is relatively large; and the emergence of MOOCs, a rising alternative, has not been seen as a serious threat to higher education institutions. There is a lack of research that has analyzed the public perception of MOOC, systematically, in a longitudinal manner with topic modelling on MOOC. This article attempts to analyze public opinion on MOOC, in both the global and Indonesian (local) context, using pretrained RoBERTa deep learning models to discover sentiment trends, and n-gram analysis, as well as topic modeling with LDA to discover themes, in order to Figure out lessons that can be learned from MOOC. It was discovered that besides a neutral sentiment, the sentiment of public opinion in both the international and local context is mainly positive. Through topic modeling, five and three main themes were discovered in the international and local context, respectively, which shows how MOOC has revolutionized the education landscape by creating a more practical, freely accessible, and highly personalized learning experience, to which higher education institutions can learn from.