Using Java for Artificial Intelligence and Intelligent Agent Systems

Intelligent Agents are being used to model simple rational behaviours in a wide range of distributed applications. In particular, agents based on the Belief-Desire-Intention (BDI) architecture have been used successfully in situations where some modelling of human reasoning and team cooperation has been needed, such as simulation of tactical decision making in air operations and command and control structures. Other applications include business process reengineering, telephone call centres, and air traffic management. However, Intelligent Agent frameworks have so far been large, monolithic software systems. With their origins in research on Distributed Artificial Intelligence, these frameworks have generally been developed as research environments in the research laboratory. Consequently they have been unduly large, complex to use and based on non-mainstream AI languages. The JACK Intelligent Agents framework presented in this paper brings the concept of intelligent agents into the mainstream of software engineering and Java. JACK is a third generation agent framework, designed to be a set of lightweight components with high performance and strong typing. We discuss the advantages and issues of using Java to implement such an Intelligent Agent framework. We present JACK’s extensions to the Java language for defining the extra concepts needed in for Intelligent Agents. We discuss the benefits of our component based approach, both for experts in artificial intelligence (such as the availability of an ever increasing amount of commercial, industrial-strength software) and the software engineer developing sophisticated distributed applications (such as n-tier business systems).