Sentiment Analysis and Topic Modeling of Tourist Reviews on Bali Island Attractions on Tripadvisor Using Lexicon-Based Method and Latent Dirichlet Allocation (LDA)

One sought-after type of information by internet users is related to tourist destinations. Hence, the need for information retrieval about a particular tourist spot they plan to visit. This study aims to analyze sentiments and identify the topics in the visitor reviews of Bali Island tourist attractions on TripAdvisor using Lexicon-based and Latent Dirichlet Allocation (LDA) methods. The data used for analysis consists of reviews from various tourist destinations on the island of Bali. For sentiment analysis, the author employs a Lexicon-based approach, focusing on both positive and negative sentiments. To identify the topics in the reviews, the author employs the LDA method to uncover the most frequently discussed topics. From 15,827 dataset, It is found that 87,6% of the responses are positive, 7.9% are negative, and the remaining 4.4% are neutral. As for the topic modeling results, the study identifies four main topics with the best coherence values based on the validation of topics with topic coherence. These four topics are: the first topic discusses experiences in Safari or Safari Park in Bali, the second topic talks about experiences in tourism in Kintamani, Bali, the third topic focuses on experiences in tourism in Nusa Penida, Bali, and the last topic discusses experiences in Scuba Diving activities