Dynamic game adaptation based on detection of behavioral patterns in the player learning curve

In educational serious games, the learning curve of a player represents his/her progress in acquiring cognitive abilities and new knowledge necessary for solving the game challenges. Hence, it is very important for an adaptive serious game to have a mechanism for detection of patterns of player´s learning curve in playing time. The paper presents the application of a game adaptation method based on automatic and dynamically detection of specific learning curves at runtime within a 3D video game of car driving in various weather conditions. The method uses a client-side software component called “Player-centric rule-and-pattern-based adaptation asset” and developed within the scope of the RAGE (Realising and Applied Gaming Ecosystem) H2020 project. The component is incorporated into a 3D car driving video game in order to enable a dynamic detection of different patterns of player performance. It allows the video game scenario to be adapted to each player by providing appropriately for him/her challenges and game features. We carried out a practical experiment with students from Sofia University, Bulgaria, where we found that adaptation of game difficulty by applying the RAGE software component resulted to improved game playability.