The importance of Diversity in Profile-based recommendations: A Case Study in Tourism

The paper explores the concept of similarity between two users measured in user profile space rather than the traditional rating space. The study aims at discovering the most relevant user profiles in order to provide recommendations to any given target profile. Closer profiles were found to be the most accurate in terms of prediction error. However, the best results were obtained when including profiles far away from the target one. This striking result is explained in light of the diversity prediction theorem.

keywords: User profile, Diversity Prediction Theorem, tourism, gastronomy