Modelling Goal-directed Players in Digital Games

David Thue and Vadim Bulitko

The pursuit of a viable model of player behaviour has gained momentum in research in recent years, and it is beginning to attract the attention of the designers of next-generation digital games. In this paper, we present a novel enhancement to player modelling that is well-suited to the digital role-playing and puzzle game industries, titled Goal-Directed- Player Modelling, in which state abstraction based on a player’s goals is used to improve the performance of a classifier for predicting player actions. We survey a set of related research, formally introduce a method for Goal-Directed-Player Modelling, and present empirical results which clearly show the ability of Goal-Directed-Player Modelling to greatly improve the accuracy of a simple, online, low cost incremental classi ier to a level near those of more advanced and complex of line methods.

This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.