Proposing an Automatic System for Generating Super Mario Bros Datasets

Abstract

Video games have become a critical part of the entertainment industry. In addition to being designed for fun, games can also be used in education, health, and many other areas. For example, some methods from the Procedural Content Generation (PCG) approach are used to create content for Super Mario Bros (SMB) algorithmically seeking to develop personalized levels based on players’ mechanics. This work aims to provide an automatic system for generating datasets composed of game logs and video frames in the Mario AI Framework, which is expected to speed up the development and validation of PCG, player modeling/profiling, and knowledge tracing methods. This also helps in the reproducibility of researches using the SMB game.