An analysis of TripAdvisor reviews of top tourist attractions in Antalya, Turkey: a comprehensive analysis

The primary objective of this research is to formulate a methodology tailored for the analysis of online reviews through machine learning paradigms, ensuring that tourism professionals and destination managers can easily interpret and act upon the findings to bolster their attractions. By harnessing data science methodologies, this research explores into both the sentiment and thematic nuances of reviews, painting a comprehensive tableau of tourist impressions and experiences. The locus of this inquiry is the reviews on TripAdvisor, focusing on Antalya, Turkey’s diverse attractions ranging from its pristine beaches and historical edifices to bustling markets and prominent cultural touchpoints. Through a rigorous analysis of 86,478 online reviews, the study employs two machine learning methodologies: latent Dirichlet allocation (LDA) and support vector machine (SVM) modelling. LDA assists in discerning the specific dimensions characteristic to each attraction type. The research elucidated four dimensions for beaches, three for ancient sites, and two each for markets and other cultural heritage sites. Additionally, this study introduces two pragmatic analytical tools, namely dimensional salience-valence analysis (DSVA) and lexical salience-valence analysis (LSVA). The findings not only underscore Antalya’s standing as a favoured tourist locale but also illuminate areas warranting enhancement, presenting invaluable insights for regional tourism stakeholders. In a broader spectrum, this research augments the burgeoning literature by underscoring online review platforms as pivotal reservoirs for gleaning insights into consumer behaviour and refining destination marketing strategies.