RPG Creature Design: Cross-System Analysis and Conversion

• Artur de Oliveira da Rocha Franco UFC
• Thiago Iachiley Araújo De Souza UFC
• Paulo de Tarso Passos da Costa Filho UFC
• José Wellington Franco Da Silva UFC
• José Gilvan Rodrigues Maia UFC
• Miguel Franklin De Castro UFC

Resumo

The Role-Playing Game (RPG) genre has achieved tremendous popularity, which is evidenced by its significant sales and high profits. One of the key reasons behind its success is the ability to model a vast array of entities, including creatures and characters, within RPG systems. However, it is worth noting that only a limited number of literary works delve into the automated conversion of entity designs across different RPG systems. This paper showcases a series of research endeavors aimed at automating RPG content, particularly focusing on creatures, involving constructing a creature database for three RPG systems, emphasizing data linkage. Three system conversion approaches are introduced, two leveraging Machine Learning (ML) techniques, and one relying on manual modeling. To assess the effectiveness of the conversions, error metrics were employed alongside feedback obtained from 14 players. Results suggest that despite ML methods yielding lower numeric errors, the creature sheets created using manual modeling were better accepted by human evaluators.

Palavras-chave: RPG, datasets, game design, machine learning, statistical analyses