Text-based and model-based process descriptions have their own particular strengths and, as such, appeal to different stakeholders. For this reason, it is not unusual to find within an organization descriptions of the same business processes in both modes. When considering that hundreds of such descriptions may be in use in a particular organization by dozens of people, using a variety of editors, there is a clear risk that such models become misaligned. To reduce the time and effort needed to repair such situations, this paper presents the first approach to automatically identify inconsistencies between a process model and a corresponding textual description. Our approach leverages natural language processing techniques to identify cases where the two process representations describe activities in different orders, as well as model activities that are missing from the textual description. A quantitative evaluation with 46 real-life model-text pairs demonstrates that our approach allows users to quickly and effectively identify those descriptions in a process repository that are inconsistent.
|Title of host publication||Business Process Management (13th International Conference, BPM 2015, Innsbruck, Austria, August 31-September 3, 2015)|
|Place of Publication||Berlin|
|Publication status||Published - 2015|
|Event||Business Process Management - Berlin|
Duration: 1 Jan 2015 → 1 Jan 2015
|Conference||Business Process Management|
|Period||1/01/15 → 1/01/15|