Research on Sentiment Classification of MOOC User Comments Based on Machine Learning

The increasing use of information technology in education has led to the emergence of online teaching methods, such as MOOC, that plays a vital role in promoting the reform and development of basic education. MOOC primarily uses microlessons to present learning content, and students can collaborate, transfer and share information by exchanging questions in the comment section. MOOC comments reflect the learners' evaluation and attitude towards the course. However, comment text is unstructured data, so obtaining accurate information about learners' emotional feedback from comment text is challenging for both users and platforms, and improving MOOC lessons using feedback from comments proves difficult. In this study, we used the "Chinese University MOOC" platform as a case study, utilizing artificial labeling to create experimental datasets. We compared the sentiment classification of four conventional machine learning models, explored error cases, and proposed a model optimization scheme. The results serve as a reference for future research on sentiment classification of spoken text data.