Title: Assessing the emotional feedback of teaching and learning service beneficiaries using machine learning on text comments

Authors: Wipawan Buathong; Pita Jarupunphol

Addresses: Department of Digital Technology, Phuket Rajabhat University, Phuket, Thailand ' Department of Digital Technology, Phuket Rajabhat University, Phuket, Thailand

Abstract: This study evaluates emotional responses to educational services using advanced machine-learning techniques to categorise sentiment in feedback. The dataset includes 1,033 comments from 402 individuals, collected via various platforms. Three algorithms were applied: random forest, Naïve Bayes, and long short-term memory (LSTM). The ten-folds cross-validation method ensured model robustness. Random forest achieved the highest F1-score of 0.833, LSTM at 0.827, and Naïve Bayes at 0.807. The analysis indicated that neutral sentiments were most accurately predicted, followed by positive and negative sentiments. Additionally, latent Dirichlet allocation (LDA) identified key themes within the feedback. Positive topics included teaching effectiveness, subject variety, and professional development. Negative topics highlighted issues with technology and resources. Word cloud dashboards focused on curriculum design, learning support mechanisms, and instructional quality. These insights are crucial for enhancing the effectiveness of teaching services, indicating areas of strength and potential improvement.

Keywords: sentiment analysis; machine learning; educational services; natural language processing; data visualisation.

DOI: 10.1504/IJIL.2025.147982

International Journal of Innovation and Learning, 2025 Vol.38 No.1, pp.22 - 49

Received: 27 Feb 2024
Accepted: 30 May 2024
Published online: 12 Aug 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article