GameRecs: Video Games Group Recommendations

Abstract

Video games are a relatively new form of entertainment that has been rapidly gaining popularity in recent years. The number of video games available to users is huge and constantly growing, and thus it can be a daunting task to search for new ones to play. Given that some games are designed to be played together as a group, finding games suitable for the whole group can be even more challenging. To counter this problem, we propose a content-based video game recommender system, GameRecs, which works on open data gathered from Steam, a popular digital distribution platform. GameRecs is capable of producing both user profiles based on Steam’s user data, as well as video game recommendations for those profiles. It generates group recommendations by exploiting lists aggregation methods, and focus on providing suggestions that exhibit some diversity by using a k-means clustering-based approach. We have evaluated the usability of GameRecs in terms of the user profile generation and the produced video game recommendations, both for single users and for groups. For group recommendations we compared two recommendation aggregation methods, Borda Count and Least Misery method. For diversity evaluation we compared results with and without the proposed k-means clustering method.