Serious games are now an increasingly used tool in business training. The question of the effectiveness of such devices on learning is a research issue. The indicators provided at the end of a video game are insufficient to understand and follow the path of a learning player. It is therefore necessary to track not only the player's actions but also to provide tools in order to analyze and diagnose the knowledge acquisition of the learner. We developed an approach based on Petri nets, used to model the accurate behavior of the player. We complete this tool with ontology to explain learner's mistakes.