GRSPOI: A Point-of-Interest Recommender Systems for Groups Using Diversification

Abstract

Context: With the massive availability and usage of the Internet, the search for Points of Interest is becoming an arduous task. Thus, Points of Interest Recommender Systems arise to help users in the search. These systems traditionally recommend points of interest to individual users, however, there are scenarios in which individuals gather, therefore creating the need to recommend items to groups. Problem: The problem is that users’ location is not always considered, only their preferences. Hence, there are studies indicating the greater is users commuting, the less POIs relevance appears to them. Furthermore, the recommendations belong to the same category, without diversity. Solution: Develop a Points of Interest Recommendation System for a group using a diversity algorithm, based on members’ preferences and their locations. IS Theory: This work was conceived in the light of the General Theory of Systems, in particular open systems as they undergo interactions with the environment where they can be inserted. Recommender systems depend on a continuous exchange of information with the external environment. Method: The research is based on the literature, and its evaluation was carried out through an online experiment with real users. The analysis of the results was carried out with a qualitative approach. Summary of Results: Precision metrics were used in the evaluation, and it was observed that the level at which the results are analyzed is relevant. For the top-3, recommendations without diversity performed better, but at the top-5 and top-10 levels, diversification had a positive impact on the results. Contributions and Impact in the IS area: A recommendation system for groups that considers the geographic location of users, their preferences and the diversity of recommendations. In addition, we provide the community with a dataset with user ratings of points of interest and geolocation information.