Sentiment Analysis of Learners’ Comments

Abstract

The rapid expansion of online learning platforms has enabled students to express themselves freely through comments, offering valuable insights into their educational experiences. The objective of this work is to propose an automatic sentiment analysis approach applied to learners’ comments, in order to determine whether the expressed sentiments are positive, negative, or neutral. To achieve this, a preprocessing phase was carried out, including text cleaning, tokenization, and automatic labeling of comments using the TextBlob library. The annotated data was then used to train an LSTM (Long Short-Term Memory) model, a recurrent neural network architecture particularly well suited for natural language processing. The proposed approach was validated using a dataset collected from the Mark My Professor platform, which contains over 5,200 reviews written by learners about their courses and instructors. Mark My Professor is a platform dedicated to the evaluation of teachers and educational content by students. The proposed model was evaluated using performance metrics such as accuracy, confusion matrix, recall, and F1 score. The results obtained show promising potential, but improvements are still needed, particularly due to the data imbalance that may have affected the model’s learning process.