A Framework to Incorporate Goals and Behavior in Supply Chain Games

Supply chain games have become an important role to teach students and managers, who can make decisions in supply chains, about supply chain management. Supply chain games should be able to capture the complexity and dynamic characteristics of supply chain management in order to give better trainings or educations to their users. Each supply chain actors has its goals and behavior that often conflict with other actors goals and behavior. Supply chain games, however, often ignore to incorporate goals and behavior in their computer-controlled actors. One of the reasons is that they focus only on certain aspects in supply chain management, such as the bullwhip effect or simple distribution management. The main challenge in this research is to support supply chain games developers to flexibly incorporate goals and behavior in their computer-controlled actors. Flexible in this research means that goals and behavior included in supply chain games depends on the games purpose or scope. This research tries therefore to build a framework to answer the challenge. This research should answer what the requirements for the framework are, what theoretical concept that can be used as the basis of the framework, and how the framework looks like. The Belief-Desire-Intention (BDI) model, which is one of the most famous models to build software agents, helps the framework to be potential in fulfilling its requirements. The BDI model provides the components, such as goal, desire, plan, and intention; and how they connect with each other. Those are useful in designing and implementing the framework. The framework is implemented and tested using a supply chain game called the Distributor Game. Some scenarios are built and some experiments are conducted to test the scenarios. The results show the potentials of the framework to connect goals and behavior, while supporting the Distributor Game developers to change goals and behavior used by their computer-controlled actors. The framework, furthermore, supports learning processes for computer-controlled actors by incorporating the idea behind the dynamic scripting. The framework is at the starting phase. Many challenges lie ahead to make the framework more sophisticated and more useful in supporting supply chain games developers, such as to include beliefs and to look at other agent models. For more information Chantal Schoof (cschoof@rsm.nl)