An approach to recommendation systems oriented towards the perspective of tourist experiences

The Community-Contributed Geotagged Photos has widely contributed to the construction of tourism recommendation systems that facilitate the task of choosing points of interest (POIs) to visit, organizing itineraries, managing activities and improving tourist experiences when they are visiting an unfamiliar city. Tourists have driven by their aspirations, desires, and preferences. There is a new kind of traveler who instead to explore the most popular places in a city; they would like to have contact with local people and culture, exploring areas where the local people usually visit considering temporal issues, such as: parts of day, day of week, vacations, events, holidays, etc. This work presents the new recommendation model, a tourist recommender system that makes POIs recommendations considering the different interaction of tourists and residents around a city over time. This new model was constructed using Naive Bayes classifier and Collaborative Filtering. The experiments used data of 83.302 photos taken in the city of Rio de Janeiro published on Flickr and 242 locations identified through OpenStreetMap. The results demonstrated that this approach could make predictions based on the temporal context and considering differences between resident and tourist perspective being this work one of the first to consider the data yielded by residents like relevant to the build of recommendations.