How to Speak 'Winese': Learning the Language of Wine Reviews

Abstract

We propose a framework integrating text mining and psychometrics to translate online product reviews into a brand positioning map that isolates their underlying quality perceptions and preferences for competing alternatives from the reviewers’ language propensities. More specifically, we develop a netnography-based product-specific vocabulary taxonomy by ontology learning. Then, we generate a perceptual map that reflects the underlying quality of the products being reviewed after adjusting for idiosyncratic linguistic differences among reviewers. The final output of our procedure is a brand positioning map similar to those that have been commonly used by marketing researchers and managers. The key difference from our approach is that our brand positioning map is constructed from consumers’ (or experts’) own consumption vocabulary in the form of online product reviews without researchers determining a priori product attributes to be used by consumers in rating competing brands. Lastly, whereas we use wine reviews produced by wine connoisseurs in our empirical application, our framework can be easily applied to a wide variety of cases where multiple consumers post their own product reviews.