Computational Issues in Modeling User Behavior in Serious Games

The main challenge when modeling user in-game behavior in serious games is to accurately infer the users’ knowledge, skills, and attributes (KSAs) from their actions in a game. Since games present a highly nonlinear assessment or learning experience, an accurate validation, not just of the resulting computational models, but more importantly, of the process and assumptions used to generate these models is needed. There are many possible issues and pitfalls in this process. Several recommendations to address these issues are given. The accurate and valid modeling of human in-game behavior provides valuable information about the game properties, the user KSAs, and their interaction in a given context or environment. When analyzing the issues of user in-game behaviors we will focus on games rather than on simulations, since we see games as extensions of simulations where simulated real or imaginary environments are extended by incentive and reward structures that trigger human behavior and responses. Such extensions are capable of producing engagement as re?ected in the success of the game industry.