Enterprise Modelling and Information Systems Architectures

Abstract

Quality assurance is a serious issue for large-scale process modelling initiatives. While formal control flow analysis has been extensively studied in prior research, there is little work on how the textual content of a process model and its activity labels can be systematically analysed. In this context, it is a major challenge to systematically identify and to consequently assure high label quality. As many large process model collections contain more than thousand models, each including several activity labels, there is a strong need for an automatic detection of labels that might be of bad quality. Recent research has shown that different grammatical styles correlate with potential ambiguity of a label. In this paper, we propose an algorithm for recognition of activity labeling styles. The developed algorithm exploits natural language processing techniques, e.g., part of speech tagging and analysis of the grammatical structure. We also study how ontologies, like WordNet, can support the solution. We conduct a thorough evaluation of the developed techniques utilising about 6,000 activity labels from the SAP Reference Model. The evaluation of this algorithm shows that spurious labels can be identified with a significant level of precision and recall. In this way, our approach can be used as a means of quality assurance for process repository management by listing bad quality labels, which a human modeler should correct.