Applying Triple Graph Grammars For Pattern-Based Workflow Model Transformations.

Abstract

Workflow and business process modeling approaches have become essential for designing service collaborations when developing SOA-based systems. To derive actual executable business process descriptions from the high-level workflow models, model transformation techniques can be used. Various service composition and business process languages are available for describing the executable processes. They have been developed having slightly different aims and requirements in mind. They do, however, share common key constructs, called workflow patterns that recur in descriptions given in these languages. We propose a model-driven approach for transforming workflow models given as UML activity diagrams into service composition descriptions. This paper will show how to realize a transformation from UML to BPEL and XPDL with a technology based on Triple Graph Grammars (TGGs). TGGs allow structural relationships between the different model elements to be elegantly expressed in graphical, declarative rules. We will show, in particular, how the commonly known workflow patterns recurring in the different business process languages can act as a guideline for designing the transformation rules. Based on the experiences in this application domain, we furthermore outline ways to enhance the usability and applicability of TGG for this purpose.