Sentiment Analysis of ChatGPT Reviews on Play Store using Support Vector Machine
Abstract
User reviews on Google Play Store contain opinions that can be used to evaluate digital applications, but large review volumes make manual analysis inefficient. This study analyzes 8,000 ChatGPT reviews collected from January to May 2026 using the CRISP-DM framework. Text preparation includes case folding, cleaning, tokenization, stopword removal, normalization, and stemming. Sentiment labels are assigned with a lexicon-based approach into positive, negative, and neutral classes, while TF-IDF converts the processed text into numerical features. The dataset is divided into 80% training data and 20% testing data, and classification is performed using a linear-kernel Support Vector Machine. Lexicon labeling yields 48.2% positive, 42.9% negative, and 8.9% neutral reviews. On 1,600 test reviews, the model achieves 81.12% accuracy. F1-scores are 0.90 for positive, 0.85 for negative, and 0.36 for neutral sentiment. The results indicate that the model performs well for positive and negative reviews but remains limited in recognizing neutral sentiment. A Streamlit interface is also implemented to present processing stages, evaluation results, sentiment visualization, and search functionality.