Extracting revisit intentions from social media big data: a rule-based classification model

Abstract

Purpose The examination of revisit intentions in hospitality is integral to relationship marketing and customer loyalty. Its measurement and determination have largely been done through closed-ended measures in surveys of customers. However, vast troves of consumer-generated media in the form of open-ended text reviews can also serve as sources for the determination of revisit intentions. The purpose of this paper is to develop and test a rule-based classification model from big data to extract revisit intentions. Design/methodology/approach Data for this came from 116,241 reviews scraped from Tripadvisor.com using a stratified sampling technique comprising hotels in major cities in the USA. A sample comprising 1,800 reviews was randomly drawn from this larger pool of reviews and manually annotated. A manual-set rule-based model, supervised machine learning (ML) models and hybrid models were developed to extract revisit intention. Findings The hybrid model of the MSRB method complemented by the gradient boosting ML method performed the best to classify revisit intentions in reviews. Practical implications This study’s rule-based classification model can be used by hotels to evaluate revisit intentions from the ever-growing pool of consumer-generated reviews. This can enable hotels to identify drivers of re-patronage and enhance relationship marketing initiatives. Originality/value This study is the first to propose an analytical model that taps big data to extracting revisit intentions. In the past, revisit intentions have been assessed using closed-ended questions using traditional survey-based methods.