The Model Morphing Approach - Horizontal Transformation of Business Process Models

By Marion Murzek .
Advised by Christian Huemer and Gerti Kappel

This work has been finished in February 2008.

Owing to company mergers and business to business interoperability, there is a need for model transformation in the area of business process modeling to facilitate model integration and model synchronisation. This need arises, on one hand, from the fact that there are many different business process modeling formalisms, for example the ADONIS R Standard Modeling Method , UML 2.1 Activity Diagram, the Event-driven Process Chain Method, and, the Business Process Modeling Notation. These formalisms provide different ways to express and represent the same aspects of business process modeling. On the other hand, existing model transformation approaches, like ATL, QVT, and Fujaba, use very general concepts for transforming models for different purposes. However, recurring structures have been observed when transforming models in the area of business process modeling. This leads to the assumption, that there are similar transformation problems in a distinct area. These recurring structures, however, are only inadequately supported by existing transformation approaches.

This thesis analyzes the different ways of how business process modeling aspects are represented in various business process modeling formalisms. Furthermore, existing transformation approaches are evaluated concerning their suitability for transforming models in the area of business process modeling. Based on this evaluation, special requirements and solutions for model transformations in the area of business process modeling are derived. These solutions lead to the construction of the Model Morphing approach, which consists of an integrated metamodel and morphing methods which operate based on this metamodel. The Model Morphing approach makes it possible to concentrate on the specific transformation problems within a distinct domain. Furthermore, it reuses existing model transformation approaches and reduces the need for excellent programming skills when defining model transformations.


Download the paper