Dissertations, Theses, and Capstone Projects

Date of Degree

9-2024

Document Type

Dissertation

Degree Name

Ph.D.

Program

Computer Science

Advisor

Robert Haralick

Committee Members

Candido Cabo

Hui Chen

Lou D'Alotto

Dov Dori

Subject Categories

Computer Sciences

Keywords

Dynamic Difficulty Adjustment, Naive Bayes, Data Decay, Feature Reduction

Abstract

Static difficulty adjustment has been applied to video games since their inception. However, dynamic difficulty adjustment did not become a topic of interest in either the academic fields or the industry until the turn of the century with sufficient advancement in processing power of computers and console systems. Amongst the work done in this area, most of the focus has either been placed on the action/adventure or the strategy game genre. However, there are only a limited number of studies regarding the role playing game genre which, by the nature of such games, generates a massive amount of data regarding player behavior.

This dissertation presents a method which makes use of the history of the player’s decisions for the purpose of moderating the performance level of the computer characters to achieve the task of dynamic difficulty adjustment in role playing games. The algorithm is designed such that it requires only a limited amount of resources both in memory and in processing power. This method is tested through a repeated simulation of battles between player characters utilizing predefined strategies that shift as time progresses against computer characters using the method presented in this dissertation. The result of the experiment shows that the method is adaptable to changes in player behavior and can perform the task of dynamic difficulty adjustment at a reasonably successful level.

Download

Share

COinS