A Pattern Based Approach for Automatic Business Process Specialization

Organizations build information systems to support their business processes. Some of these business processes are industry or organization-specific, but most are common to many industries and are used as is, modulo a few modifications. Our work tries to capitalize on these similarities to develop a methodology and tools that help business analysts generate organization-specific process models from a catalog of generic business processes. We developed a framework for representing and classifying business processes that supports process variability management by, 1) navigating a repository of generic processes, and 2) automatically generating new process variants around key process variation points. We use business patterns from the Resource Event Agent ontology to identify variation points, and to codify the model transformations inherent in the generation of the process variants. We developed a prototype, showing the computational feasibility of the approach, and validated the relevance of the variation points, and the correctness of corresponding transformations in the context of ERP key processes, showing the conceptual soundness of the approach.