Sentiment Analysis of Tourists’ Perceptions of Ubud as a World Gastronomy Destination Using the Lexicon-SVM Method

Abstract

The development of Ubud as a sustainable gastronomic tourism destination requires understanding tourist perceptions expressed on digital platforms. This study analyzes tourist sentiment toward Ubud’s gastronomy using English-language reviews from TripAdvisor and X (formerly Twitter) through a hybrid Lexicon–Support Vector Machine (SVM) approach. A total of 28,550 pieces of textual data were analyzed, consisting of 23,647 TripAdvisor reviews and 4,903 X posts. The methodology includes data collection, text preprocessing, sentiment labeling using the VADER lexicon, TF-IDF feature extraction, and SVM classification. Model performance was evaluated using accuracy, precision, recall, and F1-score. The results show that positive sentiment dominates on both platforms, with accuracies of 89.7% for X and 92.31% for TripAdvisor. Word cloud analysis further indicates that tourist perceptions are influenced by food quality, service, atmosphere, and pricing. These findings demonstrate the potential of the hybrid Lexicon–SVM approach for supporting sustainable gastronomic tourism development in Ubud. The study also contributes comparative insights into sentiment characteristics between structured reviews and real-time social media platforms.