Abstract
By using a human-centric approach to online recommender systems, this research aims to estimate the language discrepancies of which travelers and destination marketers describe the travel experiences across 11 tourism destinations in USA. In order to address the research purpose, data has been collected from two different sources that reflect the views of travelers and service providers. Then, a set of text data mining methods (i.e., clustering analysis and Jaccard distance score) was applied to identify the language differences between travelers and CVB websites, according to the following categories: shopping, dining, nightlife/activities, and attractions. Some possible methodological extensions that can improve recommendation capabilities, and managerial implications of these findings are provided.