Emergent Personalised Content in Video Games

Abstract

Games are interactive experiences by their very nature. This interactivity allows different players to tackle the proposed challenges in their own way. Most games tend, however, to provide static content that all users must experience to progress, without taking into considerations underlying preferences and playing patterns that might be unique to each player.

In this thesis, we explore a general framework that can be used to adapt the game content to the user dynamically. We present an approach that analyses in-game player behaviour and uses this data, combined with established psychological models, to generate procedural, personalized levels, to provide a more engaging and enjoyable experience.

Three parts comprise this thesis: the development of the procedural algorithm that adapts to different players, the implementation of a simple video game to test our results, and a study to collect data and validate the system. The study shows that catering to different users and adapting the game increases the average level of enjoyment.