Organizations increasingly determine process models to support documentation and redesign of workflows. In various situations correspondences between activities of different process models have to be found. The challenge is to find a similarity measure to identify similar activities in different process models. Current matching techniques predominantly consider lexical matching based on a comparison of activity labels and 1-to-1-matchings. However, label based matching probably fails, e.g., when modellers use different vocabulary or model activities at different levels of granularity. That is why we extend existing methods to compute candidate sets for N-to-M-matchings based on power-sets of nodes. Therefore, we impose higher demands on process models as we do not only consider labels, but also involved actors, data objects and the order of appearing. This information is used to identify similarities in process models that use different vocabulary and are modelled at different levels of granularity.
展开▼
