[1P4-GS-7] Agents: Game and collaboration

[1P4-GS-7-02] Evaluation of Game Scenario Generated Using Multi-Agent System by Semantic Score Method

〇Yoto Yamazaki¹, Ei-Ichi Osawa² (1. Future University Hakodate Graduate School, 2. Future University Hakodate)

Keywords:Multi-Agent system, Semantic Score , Story Generation

In this paper, we apply the semantic score method to the penetrated game scenarios using a multi-agent system. The semantic score method is a technique to analyze the structure of a movie. It evaluates the scene, the unit of the story, on a scale of complexity and solution. A semantic graph can be obtained in which the vertical axis indicates the complexity between the evaluated semantic scores and the horizontal axis indicates the scene number. Characters in the game are modeled as agents with elements such as roles, relationships, and emotions. The agents interact with each other in simulations and the multi-agent system generate a short game scenario. Behaviors of the agent are scored for complexity and solution, and we verify the semantic graph that is produced. As a result, the mountain-shaped semantic graph is generated in the game scenario generation system based on the multi-agent system.