Reading Between the Lines – Towards an Algorithm Exploiting In-game Behaviors to Learn Preferences in Gameful Systems

Players’ retainment can be fostered by investigating whether the game elements players are interacting with are to their liking and tailoring game dynamics to meet their preferences. Thus, adaptive gameplay is a widely interesting topic in both the Game User Research field and the game industry. Considering that explicit information on players’ preferences often lacks, alternative approaches are needed. This task becomes even more challenging when the gameplay data available is limited due to the simplicity of the system employed, as it occurs in gameful systems in contrast to complex entertainment games or serious games. In this work, we propose an algorithm that exploits user behaviors as an implicit component to compute players’ preferences by measuring their level of activity. The application domain is a persuasive gameful system, and the customizable game elements are single-player challenges. The proposed algorithm uses offline gameplay data to compute a preference score for every viable option. The outcomes are then compared against a ground truth calculated from players’ in-game choices. Our findings suggest that players’ behaviors can be used to inform the generation of tailored game elements.