Skip to Main Content
Modeling and implementing auction systems using agent technology is a common practice because agents can assume various roles and their behavior will be determined as a result of negotiation. However, emergent behavior is a hurdle. Mechanisms must be in place to make sure that agents participating in the auction systems won't behave in an unintended way. Detecting emergent behaviors in the design phase rather than the deployment is more cost and effort efficient. Patterns of interaction, called scenarios, are the basic modeling constructs for design and behavioral modeling of agents. However, working with several agents in an online auction system needs large number of scenarios. Therefore transforming scenarios to finite state machines (FSM) and parallel execution of the FSMs in the behavioral synthesis phase may lead to computational overload. So far all the research has been around the ways of detecting emergent behavior and scalability of behavioral modeling has been an issue. In this paper a method to identify those agents that will not cause emergent behavior is introduced. Then by eliminating them from the behavioral modeling phase, the number of FSMs and their states will be reduced. The method is explained along with a case study of a realistic online auction system that has led to 33% reduction of synthesized FSMs.