Significant word items in hotel guest reviews: A feature extraction approach

Abstract

The Internet in general and social media in particular have changed the way travellers search for information, interact and communicate with companies and among their peers. Through social media, travellers are not only content consumers but also producers who provide detailed evaluations of their experiences in a destination. Given the abundance of hotel reviews, managers need cost effective, time saving and scalable systems to identify relevant customer evaluations. This text mining study presents insights into the analysis of reviews posted on TripAdvisor. A penalized Support Vector Machine identifies keywords representative for the most positive and negative reviews. These terms can be used as an early warning system by managers to efficiently monitor the online dialogue on their hotel. The benefit of the suggested approach is the selection of the most important terms regarding the weight of the term and not regarding mere term frequency.