Toward Automatic Generation of Training Systems in RPGs: Development Elements Classification and Feature Extraction

Abstract

Currently, there are various character development systems designed to attract player interest, but quantitative comparative analyses of their differences have not been sufficiently conducted. The purpose of this study is to elucidate the mechanisms of character development systems that keep players engaged. Specifically, we analyzed the skill acquisition trends common to three systems: skill trees, jobs, and monster development, and conducted a clustering analysis on parameter growth. It was found that by narrowing the units' areas of expertise and differentiating their roles towards the end of the game, player motivation is maintained. Additionally, the clustering analysis of parameters classified growth patterns into four categories. It was revealed that in the analyzed works, there are status improvement patterns specialized in agility, as well as patterns specialized in HP and attack, or special attack.