Analysis of Sentiment from an Online Learning Platform: Model Selection
Lin Bao, Piyachat Udomwong · 2023
Nowadays, online learning has been an important and effective auxiliary method for learning. Online reviews given by learners contain a ton of information. To reveal insight from the online reviews can enhance online course qualities. Sentiment analysis is proposed here to understand what learners think. Herein, this study focuses on model selection in sentiment analysis. Firstly, an explorative study was conducted to comprehend the review data. Next, different sentiment analysis models including rule-based and deep learning methods were utilized and their performance evaluated to find an optimal model. 7871 pieces of online reviews from Udemy paid courses collected from May 2017 to June 2023 were analyzed. The study finds that RoBERTa performed better on the educational review data. The results also illustrated common words with frequent occurrence given by paid online learners.