Text Mining and Sentiment Analysis of Tourist Reviews for Heritage Attractions in Anuradhapura, Sri Lanka

Abstract

In today’s digital era, online reviews and user-generated content play a crucial role in shaping travel decisions globally, offering rich yet underutilized insights into visitor experiences. However, these valuable insights into visitors’ experiences remain largely untapped and underutilized. This study investigates the application of text mining and machine learning–based sentiment classification to analyze tourist reviews of Anuradhapura’s heritage sites in Sri Lanka, aiming to provide data-driven insights for tourism management. Reviews from Google and TripAdvisor, spanning 2018–2024, were preprocessed and classified into positive, neutral, and negative sentiments using a decision tree model. The model achieved an overall accuracy of 80.85% and substantial agreement (kappa = 0.634), effectively capturing dominant positive and negative sentiments. Analysis revealed that positive reviews were influenced by aesthetic appeal, architectural significance, and spiritual engagement, while negative
sentiments reflected operational challenges, unmet expectations, and underwhelming site conditions. The study underscores the potential of automated sentiment analysis to guide heritage site management, inform strategic interventions, and enhance visitor experiences, offering a scalable methodology adaptable to other heritage destinations globally.