Toward Automatic Generation of Training Systems in RPGs Skill Classification and Feature Extraction

Abstract

In recent years, engineering research in the field of digital games has been progressing. However, little research has been conducted to systematize the way characters grow and to automatically generate a nurturing system. In this study, we clarify the mechanism of character development systems to prevent players from becoming bored, and classify and characterize the elements of character development for automatic generation of character development systems. Skills, which are one of the elements of character development in existing games, were extracted and classified into 22 types. We further divided the time series into the beginning, middle, and end of the game, and analyzed the trends in the timing of acquiring skills. Comparison of the ratios, χ-square tests, and residual analyses of the time series for each game showed that the common trends for each game were a decreasing trend for "weak assistance" and "abnormal conditions" and an increasing trend for "special attacks" and "enhanced assistance. The results of this study can be applied to the design of game systems to prevent players from becoming bored.