Abstract:

In a context of serious games learning, we propose a modeling of the player-learner based on bayesian network technology coupled with the diagram of influence. This modeling has for objective evaluating the effectiveness of the serious game in learning through the evaluation of the behavioral change of the player-learner. This approach was applied to a persuasive serious game of anti-corruption.

Introduction:
The technological progress, the digital practices of the teenagers and the expanding institutional context tend to renew the interest today carried for the games to educate and form. For the researcher, it is the question of the relations between game and learning, which is again on the agenda. The works lead in this domain concern the uses of the Serious Games (SG) in the education and training [1].
In the literature, different types of SG are presented [2]. By taking into account their main intentions, SG can be distributed in three big categories: SG of message, deliver a message in an educational, informative and persuasive aim. SG of training, improve the cognitive or driving performances of the users. SG of simulation or serious play, does not present objective to estimate the users, it offers an open panel of practices.

Persuasion and Player-Learner:
In [3], authors says that a big part of the SG are said "persuasive". This type of SG is defined as media of persuasive communication, often intended for a wide public. They incite the players to solve “playful problems” with the aim of provoking with them behavioral changes. The persuasive SG remain less studied. Whatever is the field of application, its fundamental objective is to provoke on players cognitive, affective and behavioral effects.
In the Persuasive SG, the Player-Learner (PL) is put right into the action; he is not a passive observer. That is why several researches focus on how to follow the learning progress of the PL. In main works, they need to design a model of the PL.

Modeling Player-Learner:
We propose an approach following three steps:
• Identifying the PL’s behavioral characteristics using a psychological test before beginning playing the SG. This test can reflect the initial behavior of PL.
• Treating information coming from both PL behavioral characteristics and its SG results, by mean of a model based on Bayesian Network (BN) technology and its associated mechanism of inference.
• Evaluating the PL behavior after each game session using the influence diagram.

Model application:
We applied the proposed modeling using one persuasive serious game, which makes sensitive against corruption. This anti-corruption SG presents several sessions where each session illustrates a scene reflecting real situation, i.e. in school, in hospital, etc. The aim of this SG is to change the PL behavior with regard to corruption. We, then, applied our model in order to evaluate PL behavior change and at the same time to evaluate the persuasive SG effectiveness.

References:
[1] J. Alvarez and D. Djaouti, Une taxinomie des SeriousGames dédiés au secteur de la santé, REE, N°11, 2008.
[2] L. Michaud, Serious games Advergaming, edugaming, training…, Understanding the digital world, IDATE Consulting & Research. www.idate.org. 2008.
[3] M.P. Fourquet-Courbet and D. Courbet, Les serious games, dispositifs de communication persuasive. La Découverte « Réseaux » 2015/6 N°194, pages 199-228. ISSN 0751-7971. ISBN 9782707188168.

Keywords:

Serious Game, Bayesian Network, Inference diagram, Behavior evaluation.