A Human-Centered Automated Framework for Integrating AI in Balancing Numerical Generation for Collectible Card Games

Collectible card games (CCGs) differ from traditional card games by requiring strategic deck-building with unique cards defined by numerical attributes like cost, attack, and health. Balancing these attributes is complex and typically relies on resource-heavy manual adjustments and testing. To address this issue, we propose a human-centered automated framework designed to generate balanced numerical attributes for new CCG cards. Our framework incorporates existing cards to produce new abilities and calculates balanced numerical attributes through an innovative evaluation metric based on reward distributions. Experimental validation demonstrates the framework's superior efficiency and effectiveness compared to traditional evolutionary algorithms and manual balancing methods, significantly reducing development time and enhancing game diversity and player engagement.