Mining Player Behavior Patterns from Domain-Based Spatial Abstraction in Games

Identifying explainable player strategies and decision patterns that give insights into player behavior is one of the most difficult tasks for game analytics, yet yields great informative potential for various purposes. Industrial stakeholders can capture player experience and infer issues or feedback on design, content, and game balancing - while players themselves might want to leverage this technique to contrast their style of play to other players, fostering self-regulated learning. On top of that, in the case of educational games, the identification of learning strategies (as well as the discovery of popular erroneous strategies) could even elevate their potential to successfully communicate educational concepts. To advance this field, we investigate how the spatial map of a game contributes to identifying player strategies and emphasize the importance of the appropriate level of abstraction to capture strategical decisions. Using visualizations and expert domain knowledge about spatial abstraction, we illustrate how the partitioning into affordance zones can reveal patterns and strategies in gameplay. To showcase the generalizability of our methodology, we investigate two case studies for the distinct genres of educational games (Parallel) and MMORPGs (Guild Wars 2). The two case studies unveil insightful strategies between different player sets of interest – only possible by the apt level of spatial abstraction.