Understanding Player Patterns by Combining Knowledge-Based Data Abstraction with Interactive Visualization

Abstract

Digital games are often created with large virtual worlds and a high number of degrees of freedom for players. For analyzing patterns of play, it is imperative to consider all features affecting the gameplay, which leads to an explosion in data space. To address this problem, we propose an approach for knowledge-based data abstraction -- inspired by applications in medicine -- that uses interactive visualizations based on game telemetry data to study player patterns. The approach involves iterative knowledge-based abstraction of elements of player action sequences to condense the data space to a level interpretable by game designers and user researchers. We developed this approach as part of developing a puzzle game using an existing interactive visualization tool, and demonstrate its value for understanding this game's player patterns. Based on the lessons learned from this study, we present a general set of guidelines for knowledge-based data abstraction for other game genres and interactive visualization systems.