Understanding Learners' Behaviors in Serious Games

Understanding play traces resulting from the learner’s activity in serious games is a challenged research area. Especially, when the serious game is characterized by a large state space and a large amount of free interactions, the play traces become complex and thus hard to analyze and to interpret by teachers. In this paper, we present a framework that assists designers to build a model of an expert’s solving process semi-automatically. Based on this model, we propose an algorithm that analyzes player’s traces in order to generate pedagogical labels about the learner’s behavior. We carried out an experimental study which aimed to evaluate the effectiveness of the labeling algorithm. From a usability point of view, we also use the experiment to validate the acceptation and readability of pedagogical labels by the teachers.