Studying Online Travel Reviews related to tourist attractions using NLP methods: the case of Guanajuato, Mexico

Abstract

ABSTRACT This research's main objective is to analyse Online Travel Reviews (OTRs) related to tourist attractions aiming at identifying recurring discussion topics/themes to infer whether travellers make reference to positive or negative experiences during their trips. This study focuses on the case of Guanajuato, a Mexican cultural destination. It is important to mention that similar analyses have not been carried out to study any Mexican destination so far. OTRs are analysed using two Natural Language Processing approaches: Mutual Information Ranking and Jaccard Coefficient. These are applied to quantify and extract the most representative themes and the main topics from each polarity within the OTRs. These techniques have been extensively studied in various areas, however, to the best of our knowledge, they have not been applied in digital tourism to detect essential issues during the travel experience. As a result, two recurrent negative themes/topics (‘cleanliness’, ‘prices’) were identified throughout this analysis. One surprising finding of this work relates to the lack of variation between national and international travellers' evaluations. The findings of this study contribute to confirming the role of OTRs within the contemporary tourist experience by identifying those places perceived negatively or positively directy from the visitors' perspective.