Beatrice Valeri Marcos Baez Fabio Casati
In this paper we study the factors that affect people’s decision in participating to leisure activities in the social and cultural environment. To this end, we collected the ratings of local people from three different cities around the world on standard leisure activities, and looked at the personal, social and contextual features shaping their preferences. We then used this dataset to evaluate how these features can be exploited to recommend places people would actually like. Our initial results suggest that friends are a good source for recommending activities, with higher precision and recall than considering only popular activities; but these can be improved reducing the scope to similar friends in the context of the particular activity. We have also found that people preferences are sensitive to the companion (e.g., partner, friends, tourists) for which they look for different features. The results also suggest that similarities in the preferences of people can be extended to other activities, which points to the potential of profiling users based on lifestyle. We finally present the design and prototype of a system, namely ComeAlong, which aims at helping people discover, find and participate to social and leisure activities.