Token analysis of graph-oriented process models

In business process management, graph-based models are used to represent coordination protocols between collaborating partners as well as for service orchestration and composition. At runtime however, current process engines are commonly based on mainly block-structured languages, such as BPEL, that differ structurally and semantically from process graphs. Recent work has accomplished elaborate mappings between both representations. Although most mappings strongly depend on the segmentation of the graph-model into components, the necessary graph-decomposition itself is not described in these works. This paper presents a novel approach based on token analysis to automatically identify components. The technique also allows for optimizations by integrating further steps in the translation of process graphs to executable workflows.