Sentiment Analysis of Chinese Online Course Reviews Based on Deep Learning

Abstract

Purpose – Sentiment analysis of Chinese online course reviews is one of the key technologies to improve the intelligence level of online learning systems. This study proposes a sentiment analysis model for Chinese online course reviews based on the long short-term memory network (LSTM). It integrates it into the "Mushroom Community" platform independently developed by the researchers to achieve automatic classification of the sentiment tendency of reviews.

Method – Taking 29,985-course reviews from the China MOOC platform as the dataset, four models, namely linear regression, support vector machine, random forest, and LSTM, were constructed and compared. The accuracy, precision, recall, and F1-score were used to evaluate the model performance.

Results – Experimental results show that the LSTM model performs best in this task: accuracy 0.94, precision 0.83, recall 0.86, F1 Score 0.84; and has been successfully deployed in the "Mushroom Community", verifying its feasibility and practicality.

Conclusion –  This study confirms the effectiveness of the LSTM model in sentiment analysis of Chinese online course reviews and can be applied to actual online learning systems.

Recommendations – Given LSTM’s reliance on large-scale, high-quality annotated data and its “black box” characteristics, subsequent work will explore the introduction of pre-trained language models such as BERT and lightweight integration solutions in small sample scenarios.

Research Implications – The research results provide a reference for Chinese text mining and the development of intelligent online learning systems.

Practical Implications – This study can provide data support and decision-making reference for educational platforms, teaching institutions, and teachers in optimizing course quality, adjusting teaching strategies, and sustainable development of online learning systems.