Comparing structure-oriented and behavior-oriented variability modeling for workflows

Workflows exist in many different variants in order to adapt the behavior of systems to different circumstances and to arising user's needs. Variability modeling is a way of keeping track at the model level of the currently supported and used workflow variants. Variability modeling approaches for workflows address two directions: structure-oriented approaches explicitly specify the workflow variants by means of linguistic constructs, while behavior-oriented approaches define the set of all valid compositions of workflow components by means of ontological annotations and temporal logic constraints. In this paper, we describe how both structure-oriented and behavior-oriented variability modeling can be captured in an eXtreme Model-Driven Design paradigm (XMDD). We illustrate this via a concrete case (a variant-rich bioinformatics workflow realized with the jABC platform for XMDD), and we compare the two approaches in order to identify their profiles and synergies.