Systematic Literature Review of Sentiment Analysis on Various Review Platforms in the Tourism Sector

Sentiment analysis has become an essential tool for understanding public opinion, especially in the digital era. This study aims to systematically review the methods and algorithms used in sentiment analysis of reviews in the tourism sector using datasets from social media and digital platforms from 2019 to 2024. The study adopts the Systematic Literature Review (SLR) methodology based on Kitchenham's guidelines, comprising three phases: planning, execution, and reporting. Data were collected from academic databases such as Scopus, IEEE Xplore, and ScienceDirect, with inclusion criteria covering relevant articles published between 2019 and 2024 and using datasets from social media platforms like Twitter or tourism platforms like TripAdvisor. A total of 22 models and algorithms, including deep learning, machine learning, hybrid, transformer, and lexicon-based methods, were identified in this analysis. The findings indicate that the methods with the highest accuracy are lexicon-based algorithms such as VADER (accuracy of 98%) and machine learning algorithms such as the Naïve Bayes Classifier (F1-score of 96%). This study also highlights the importance of data pre-processing to improve model performance. This research provides insights into trends, strengths, and weaknesses of the algorithms used in sentiment analysis within the tourism sector, as well as recommendations for researchers and practitioners to select the most suitable methods for their needs. The results are expected to contribute to the development of more optimal sentiment analysis methods for the tourism sector.