Proposal of fellow AI that learns user's favorite behavior in RPG battle

Abstract

In RPG, players form a party with fellow characters and give instructions according to the situation and role of that character during battle. Some games allow players to set action policies for their fellow characters and let their AI automatically select actions. In these games, fellow characters AI acts according to the algorithm, but that is not always the preferred behavior for the player. In this research, we propose an fellow characters AI that learns the user's favorite behavior and reflects it in the behavior.