Online Course Quality Evaluation Based on BERT

Abstract

In order to evaluate the quality of online courses, this paper proposes a framework based on online course feature extraction and sentiment analysis, and applies this framework to the online courses of MOOC. Extract the word pair of the review data through the word frequency syntactic dependency, and merge the word pair into the sentiment classification of the BERT model to realize the fine-grained feature analysis of the online course review data, so as to obtain online courses in each Use this aspect to evaluate the quality of the course. Experiments conducted on MOOC online course reviews show that the BERT model incorporating binary features has improved accuracy, recall, and F1 values compared to traditional machine learning methods.