Sentiment analysis in hospitality and tourism: a thematic and methodological review

Purpose This study aims to conduct a systematic review and critically analyze the sentiment analysis literature in hospitality and tourism from methodological (data sets and analyzes) and thematic (topics, theories, key constructs and their relationships) perspectives. Design/methodology/approach Qualitative thematic review and quantitative systematic review were performed on 70 papers obtained from hospitality and tourism categories of two databases, namely, Web of Science and Scopus. Findings A total of 5 topics and 27 sub-topics were identified and the major theme is market intelligence. Sentiment variables were investigated not only as independent but also as dependent variables. The customer rating is the most investigated dependent variable, whereas moderators and mediators were rarely tested. Most reviewed studies did not use theory. The findings from the methodological review show that analysis of big data was rare. Moreover, testing the performance of sentiment analyzes was uncommon, and only one paper tested the performance of aspect/feature extraction. Research limitations/implications This study extends prior review studies by providing a comprehensive view of how knowledge and methodologies of sentiment analysis have developed. The identified themes and key constructs serve as a solid base for future knowledge advancement. Future research directions on sentiment analysis are also provided. Originality/value To the best of the authors' knowledge, this study is the first comprehensive methodological and thematic review of sentiment analysis in hospitality and tourism. Based on the identified findings, the authors propose several directions for future research.

Automated summary

This study conducted a systematic review and critical analysis of sentiment analysis literature in hospitality and tourism, identifying 5 main topics and 27 subtopics, with market intelligence being the major theme. The study found that most reviewed studies did not use theory and the analysis of big data was rare.