This paper describes an approach to interoperability in design projects that is based on computational agents customizing the representation of product data to individual design tools. The agents can augment their translator capabilities autonomously at runtime. This has the potential to lessen the need to manually pre-define an appropriate set of tool translators. The process of generating tailored product models uses an approach to organizing agent knowledge that limits computational complexity when large numbers of tools are involved. We report the implementation and evaluation of an agentbased system that demonstrates fundamental features of mass customized interoperability. Results include an average reduction of manual work by 88% using this approach.
Udo Kannengiesser, John S. Gero