Extending Objects to Model Agents

Abstract

Incorporating the agent paradigm into geographic information science is one approach to modeling the complex interactions between natural and human systems. In the agent paradigm, systems are abstracted to a basic structural and functional form and programmed as agents that respond to changes in the environment, interact with other agents, have control over their behavior, achieve specific goals, and change their behavior using past experiences. As a result, the agent paradigm has created a growing need for standardized tools to support agent-based design, analysis, and documentation. Since the design, analysis, and documentation are dependent on input from many types of users and modelers, an emerging challenge is how to develop and document a consistent agent-based model. Based on the similarities between agents and objects, this study examines the use of the Agent UML (AUML) extension to develop and document agent-based designs. A framework is proposed to use the AUML extension in a collaborative group process to support consensus agent designs.