Image-based Travel Recommender System for small tourist destinations

Abstract

Most tour planning systems have similar three-step structures: definition of the tourist profile, evaluation of the Points of Interest (PoI), and route optimization. Having in mind that a picture paints a thousand words, this paper describes an approach that allows tourists specify their interests through a set of images, from which the system inferences their profile. Taken into account the choices made by the visitor, the system infers a dynamic profile. Furthermore, the System calculates a list of resources that match the profile with the destination tourist resources. Each resource is associated with a weight which indicates the utility of that resource to the profile of the visitor. Thus, visitors obtain personalized tourism recommendations based on their preferences as a result.