A Component Based Agent Framework for Non-experts

A software agent is viewed as a piece of software that acts on its own and interacts with other similar entities in an environment to achieve its design goals. This agent view provides a level of abstraction well suited for modeling and building complex software systems [9]. Hence a new paradigm in software engineering, namely Agent Oriented Programming (AOP) [16] and Agent Oriented Software Engineering (AOSE) have emerged. However one of the problems with building and modifying agent systems is the expertise required to carry out these tasks. Even though numerous toolkits and methodologies exist, they require strong programming skills and conceptual knowledge in agent theories to use them. In complex environments where agent systems operate, updates are commonplace as the understanding of the system grows. Most of the time it is the domain experts who identify and require these changes. For example, a meteorologist who wants an agent in a weather alert system to respond to a new type of weather change. Therefore the need for expert agent developers hinders the evolutionary development and makes it costly to maintain agent systems. Further the notion of code reuse (as in components) is not well established within Agent-Oriented Software Engineering.