Measuring User Satisfaction Through Sentiment Analysis of Gojek Reviews in Bahasa Indonesia
Year: 2025
Type: article
Abstract: User satisfaction plays a crucial role in evaluating mobile applications, especially in highly competitive markets such as ride-hailing and digital payments. This study measures user satisfaction with the Gojek application through sentiment analysis of reviews in Bahasa Indonesia, collected from the Google Play Store. The dataset, sourced from a public repository, contains 224,981 reviews across multiple app versions. After validation and cleaning, each review was mapped into three sentiment categories: Negative 1–2 stars, Neutral 3 stars, and Positive 4–5 stars. Text preprocessing included lowercasing, punctuation and number removal, stopword filtering, and stemming with the Sastrawi library. Reviews were then transformed into numerical vectors using TF-IDF. Two machine learning models were applied: Logistic Regression and Linear Support Vector Machine (SVM), trained 80% of the data and tested on 20%. Results show that SVM achieved higher accuracy 86.37% compared to Logistic Regression 83.38%. Both models performed well in classifying positive sentiments, but SVM was more effective in handling neutral reviews. Word cloud visualization highlighted frequent positive terms such as “mantap” and “mudah”, while negative reviews emphasized “error" and “lambat”. This research demonstrates the effectiveness of sentiment analysis for measuring user satisfaction in Bahasa Indonesia. The findings provide actionable insights into user perceptions of Gojek and offer a framework for developers to identify issues and improve app performance.
Authors: Arsyad, Muhammad, Ardiansyah, Mursyid
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7216631913