Listening to Student Voices: Aspect-Based Sentiment Analysis of Academic Services Using BERT

The inability to systematically process large volumes of unstructured student feedback hinders the enhancement of academic service quality in higher education. To address this challenge, this study develops an Aspect-Based Sentiment Analysis (ABSA) model using a fine-tuned BERT architecture. applied to 1,110 student reviews at Universitas Muhammadiyah Surakarta. The model was trained and evaluated using a dataset of 1,110 student reviews, filtered from an initial dataset of over 40,000 raw data points. To assess its performance, standard metrics such as accuracy, precision, recall, and F1-score were employed. The model demonstrated high performance, achieving an overall accuracy of 98.6% and an F1-score of 0.92 for identifying service aspect terms. The analysis successfully extracted key aspects, including staff interaction, administrative processes, and service efficiency. Critically, it revealed that staff interaction was the aspect with the most significant negative sentiment, providing a clear target for institutional improvement. This research confirms that the BERT-based ABSA model is a reliable and scalable tool for transforming qualitative student feedback into actionable, data-driven insights, enabling targeted enhancements to academic service quality.