A Sentiment Analysis of the Tourist Reviews of the Attractions in Saudi Arabia Using Deep Learning Models

This paper aims to enhance the standards of tourism and tourist experiences in Saudi Arabia, thereby contributing to the achievement of Saudi Vision 2030 goals. One of these goals is to stimulate economic growth by increasing tourism-related commerce, as Saudi Arabia is becoming an increasingly popular tourist destination. This study employs sentiment analysis of tourist experiences in places recognized as popular attractions by the Saudi Ministry of Tourism. Arabic language tourist reviews of Boulevard Riyadh City, Al-Ula Old Town, the Al-Balad district in Jeddah, the Heritage Village in Dammam, and the Al-Hada cable car from Google Maps are collected, and the sentiment of the reviews is classified as positive, negative, or neutral. The textual representation techniques employed in the model word embedding methods, such as AraVec, ArBERT, Qarib, and MARBERT. Moreover, various models, including baseline models, Deep Learning (DL) models, and transformer-based models, were implemented to predict sentiment. Various metrics, including accuracy, precision, recall, F1-score, and the Area Under the Curve (AUC), were used to evaluate the models' performance. The results demonstrated that QARiB achieved the best performance.