This paper proposes an agent-based social simulation system with Organizational-Learning Oriented Classifier System. The system is characterized by the agents that (1) solve multi-objective problems, (2) pursue different goals, (3) form classes to behave and/or learn and (4) live in multiple environments. We report three organizational learning techniques, "Copy of other's actions" and "Copy of other's rules", that accelerate the evolution of intra-class agents, and "Intervention to other's actions" that accelerates co-evolution between different kinds of agents. We implement the system applied to "Ecological Marketing", which reveals interesting agent behaviors on the domain.
各種検索
サポート
ヘルプ