Publication

Abstract : Tourists are increasingly relying on social media platforms to share their impressions and ideas about various places. To improve customer satisfaction and service quality, tourism professionals must understand these sentiments. However, conventional methods of sentiment analysis often struggle to capture the subtle context and specific language of tourism content. Another factor that affects the accuracy of sentiment classification is the lack of integration of domain-specific knowledge. Using social media data, this study presents a Knowledge-Enhanced BERT (KE-BERT) model to analyze feelings about various things related to tourism destinations, with the aim of tackling the above challenges. The KE-BERT model enhances its understanding of aspect terms and sentiment within the BERT framework by incorporating knowledge from the tourism industry. A comprehensive dataset of social media reviews and comments on popular tourist destinations around the world was used. The training and evaluation of the KE-BERT model utilized several performance metrics, including accuracy, F1-score, recall, and precision. The experimental resultsshow that KE-BERT outperforms existing models, including traditional BERT- and LSTM-based models, achieving an accuracy of 92.7% and an F1 score of 91.5%.