Towards Effective Distance Education Implementation: Utilizing Descriptive Analytics, Opinion Mining, and Sentiment Analysis for Online Education Mentors and Learning Materials

Abstract

This study utilizes sentiment analysis, an application of Natural Language Processing (NLP), to automate and improve the evaluation of online student feedback. Student comments from academic years 2018–2025, were cleaned, preprocessed, and analyzed using the VADER sentiment tool to label feedback as positive, negative, or neutral. These labeled data were further used to train a neural network model that uses Long Short-Term Memory (LSTM) to improve sentiment classification. Tokenization, stopword elimination, lemmatization, and contraction handling were all parts of the preparation step. VADER proved effective in detecting sentiment polarity and intensity in short student comments, while LSTM achieved an overall accuracy of 88.6%, particularly strong in classifying positive and neutral sentiments. A confusion matrix was used to assess model performance, measuring precision, recall, and F1-score. The descriptive method of textual data analysis provided insightful information about the issues that concerned students in various Online departments. The findings underscore the value of automated sentiment analysis as a feedback tool to continuously improve mentor performance and course delivery in online learning environments. The study also highlights the need for balanced training datasets to enhance the classification of all sentiment types.