Player Segmentation with INSPECT: Revealing Systematic Behavior Differences within MMORPG and Educational Game Case Studies

Abstract

Visualization tools have aided game analytics in a multitude of ways. Yet, producing customized representations that focus on specific research questions remains an open challenge for the field, as ideal sweet spots between noisy and under-representative data are almost impossible to achieve with current automated or AI-driven approaches. Following a human-in-the-loop setup, we introduce the interactive visualization tool INSPECT that applies multiple configurable noise reduction, segmentation, and contrast functions in order to highlight systematic differences or points of interest within player behavior. With respect to self-directed learning among players in complex video games and the identification of erroneous decision making in educational games, we set up two case studies with domain experts using INSPECT that ascertained and interpreted crucial player decisions that led to optimal or inexpedient behavior.