Current translation approaches from activity-centric process models to artifact-centric Guard Stage Milestone (GSM) models operate on the syntactic level. While such translations allow equivalent traces (behaviors) of executions, we argue that they generate poor GSM models for the intended audience (including business managers and process modelers). A specific deficiency of these translations is their inability to relate to relevant domain knowledge, especially groupings of activities to achieve well-known business goals cannot be obtained by syntactic translations. Ironically, this is a main principle of GSM models. We developed an initial ontology based translation framework [14] that incorporates the missing knowledge for improved translations. In this paper we further extend this framework with two metrics for the assessment of quality aspects of resulting GSM translations with domain knowledge, propose a novel semantic rewriting algorithm that enhances the quality of GSM translations, and provide an evaluation of the achievable quality for different classes of input processes. Our evaluation shows that maximum quality scores are achievable if semantics and structure of the input processes are well aligned. Given poorly aligned input processes, a translation method can optimize one of the metrics but not both.
CITATION STYLE
Köpke, J., & Su, J. (2016). Towards quality-aware translations of activity-centric processes to guard stage milestone. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9850 LNCS, pp. 308–325). Springer Verlag. https://doi.org/10.1007/978-3-319-45348-4_18
Mendeley helps you to discover research relevant for your work.