This paper addresses automatic analysis of the meaning of business process models, and is likely to be of interest to anyone who needs to extract meaning from the process models created by organizations to support quality assurance, management, and strategy development.

Over time, companies accumulate numerous detailed models from which stakeholders need to abstract higher-level views. While well-established techniques are available to perform this abstraction, the problem of giving appropriate, memorable, meaningful names to the abstract models remains. This paper presents an automatic approach to generating a ranked list of names based on the labels in a process model. It discusses several theories of meaning, relating each of these to process names.

The paper goes on to consider label analysis and classification using sources of information indicated by the theories of meaning. Seven major algorithms that together generate process names are then presented and empirically validated using three different process model collections spanning two different natural languages (English and German). Performance was excellent with all three collections, but, more importantly, four business analysts with substantial modeling experience were positive about the approach. They also suggested the possibility of identifying inconsistent naming in models.

Further applications of the research include semantic analysis of labels to support model matching, identification of similar processes, and checking the semantic quality of process models. Overall, this paper represents a very useful contribution to business process modeling with implications for any form of modeling that incorporates natural language.