Learning agents in a business simulator

Business simulation modeling is a traditional collaborative learning tool to give students an integrated perspective to business management. Recent progress of AI technology makes it possible to build business models with human- and software-agents participation. This paper describes a novel business simulator development for educational use in marketing domains. Our approach consists of (i) gaming experiments among human players, (ii) home-made simulation models by the students themselves using a business model description language (BMDL), a business model development systems (BMDS), and corresponding decision rules of the agents, and (iii) machine learning agents with classifier systems (LCSs) to improve the decision knowledge and the game balance. This paper demonstrates the effectiveness of our approach to business simulation by employing machine learning agents.