Abstract
TripAdvisor is one of the largest travel websites. Among the provided services, it aids users with suggestions about attractions, accommodations, restaurants, etc., based on a wide system of reviews. In fact, users looking for suggestions shall sort through opinions posted by any kind of other users, possibly with different preferences and travelling behaviour. The aim of this work is to provide a personalized recommendation system to integrate the TripAdvisor services, based on the identification of similar travel products rated by similar users. Some alternative models of co-clustering are considered, to handle user rates in the form of ordinal data, and to account for missing values, due to the intrinsic fact that each user rates only a small subset of the considered travel products. Possible extensions are discussed to include additional information in the model, based on products and users characteristics.