Automatically-Generated Agent Organizations for Flexible Workflow Enactment

The use of multi-agent systems for the implementation and management of data-intensive workflows is generally considered a particularly interesting solution. We propose an approach that exploits the definition of processes described with the BPMN language for the automatic generation of agent organisations. To demonstrate the use of our approach, we chose the JaCaMo agent framework for its inherent support of agent organisations described with the MOISE meta-model. The resulting agent organization is built for adapting to different execution context, by self-modifying its structural and functional specification for continuing to fulfil the BP goal.