Abstract

Improving service quality has become a major focus for public service providers, including in the drinking water sector. Customer reviews reflect both satisfaction and direct experiences. This study aims to detect emotions in customer reviews using the IndoBERT model and compare its performance with other machine learning algorithms, such as Logistic Regression, Support Vector Machine (SVM), and Random Forest. The study follows an experimental method by applying different models to a dataset of customer reviews to evaluate and compare their performance in emotion classification. The dataset consists of 63521 customer reviews categorized into five emotion labels: "satisfied," "disappointed," "angry," "urgent," and "neutral." The analysis involves data preprocessing, tokenization using IndoBERT, feature extraction through word embeddings, and classification using machine learning algorithms. Model evaluation is based on accuracy and F1-score. The results show that IndoBERT combined with Logistic Regression and SVM performs best, reaching an accuracy of 89.08%, while Random Forest achieves only 77.91%. The study concludes that IndoBERT is an effective approach for emotion detection in customer reviews and holds strong potential to support more targeted improvements in PT Air Minum Giri Menang’s service quality and response.

Concepts :

Access to Document

SDGs

Citations by Year