Machine Learning-Based Sentiment Analysis of User-Generated Content: Insights from Google Maps Reviews

Abstract

The rapid growth of user-generated content on digital platforms has provided valuable opportunities for understanding customer perceptions through sentiment analysis. Google Maps, as one of the most widely used location-based services, allows visitors to share their experiences and opinions regarding public facilities and commercial destinations. This study aims to analyze visitor sentiment toward Living World Bali using machine learning techniques. A total of 4,060 Google Maps reviews were collected and processed through several text preprocessing stages, including cleaning, case folding, normalization, tokenization, stopword removal, and stemming. The Term Frequency–Inverse Document Frequency (TF-IDF) method was employed for feature extraction, while the Random Forest algorithm was utilized for sentiment classification. Experimental results indicate that the proposed model achieved an accuracy of 91.75%, precision of 92.39%, recall of 98.84%, and an F1-score of 95.55%. The sentiment distribution analysis revealed that positive sentiment dominated the dataset, accounting for 84.70% of all reviews, suggesting a high level of customer satisfaction with Living World Bali. Furthermore, the findings demonstrate that the integration of TF-IDF and Random Forest provides an effective approach for classifying textual reviews in the tourism and retail domains. The proposed framework offers valuable insights for shopping center management in evaluating customer experiences and supporting data-driven decision-making processes to improve service quality and visitor satisfaction.