What's Vs. How's in Online Hotel Reviews: Comparing Information Value of Content and Writing Style with Machine Learning

Writing style is an important rhetorical feature of textual information. However, its value has not yet been well understood within the context of social media. This research compares two major aspects of textual content, i.e., content and style, to determine the information value of online hotel reviews. Using TripAdvisor hotel reviews, several machine learning techniques based on natural language processing (NLP) are applied to predict review helpfulness. The results indicate that textual features are core features of online reviews; that style is a more influential aspect than content; and, that combining both features produces the best results. This study contributes to the understanding of user-generated content in the textual format within the hospitality and tourism contexts. Limitations and directions for future research are also discussed.