Let the Games Speak by Themselves: Towards Game Features Discovery Through Data-Driven Analysis and Explainable AI
Riccardo Martoglia, Matteo Pontiroli · 2021
The idea behind this work is to start exploring the application of data analytics and (explainable) machine learning techniques to better understand games and discover new features that will possibly help in effectively exploiting them in different socially useful domains. We prove the feasibility of the idea by: (i) collecting a large dataset of board game information; (ii) designing and testing an information processing pipeline for automatically discovering game categories and game mechanics, with some first encouraging results. In the future, we plan to further generalize this approach for different kinds of games and for discovering currently unknown but useful aspects, e.g. games or game features that could better foster Computational Thinking in education, those better suited to be applied in social distancing contexts, and so on.