Autonomic Business Processes

Business processes in large organisations are typically poorly understood and complex in structure. Adapting such a business process to changing internal and external conditions requires costly and time consuming investigative work and change management. In contrast autonomic systems are able to adapt to changing environments and continue to function without external intervention. Enabling business processes to adapt to changing conditions in the same way would be extremely valuable. This work investigates the potential to self-heal individual business process executions in generic business processes. Classical and Immune-inspired classification algorithms are tested for their predictive utility with Decision Trees augmented with MetaCost and Immunos 99 exhibiting the best performance respectively. An approach to deriving recovery strategies from historical process data in the absence of a process model is presented and tested for suitability. Also presented is an approach to selecting the best of the determined recovery strategies for application to a business process execution, which is then tested to determine the impact of its parameters on the quality of selected recoveries.