The modelling of operational processes for the documentation, communication, improvement, and for the business process management (BPM) plays a central role for companies of all kinds. With new and increasingly automated survey methods, such as process mining, the number of documented processes increases rapidly. Process collections by companies easily reach a range of thousands to tens of thousands of individual models, the handling of which is highly problematic, costly and resource intensive even with established modelling and business process management tools. For this reason, companies are finding it increasingly difficult to react flexibly to changes in processes and adequately take into account possible side effects.
An important starting point for addressing this challenge is the development of automated procedures for the identification of correspondences between individual process models (see figure).
Particularly challenging are the following aspects :
- Types of equivalence relations: e.g. action or goal equivalence, analogy, content consistency
- Cardinality of node correspondences: 1: 1, 1: N, N: M
- Type Conflicts: degrees of freedom in modelling can lead to type conflicts in correspondence relations
The knowledge of these correspondences allows for a multitude of new scenarios for the analysis and continuous improvement of business processes. Examples include the inductive development of reference processes, support for the implementation of legal regulations, or the standardization of internal business processes. At the same time, this knowledge is essential for the flexible response to external or internal influences, because only then can the scope of these influences be reliably estimated.
Against this background, the Process Model Matching Contest deals with the analysis of existing procedures for the automatic identification of these correspondences and in particular with the assessment of their performance in practical application. After the RefMod-Miner team received the Outstanding Matcher Award in the first contest 2013 in Beijing  with the RefMod-Mine / NSCM algorithm, the second contest followed at the BPM 2015 in Innsbruck . Important innovations were the qualitative revision of the existing evaluation data records as well as the introduction of new data records from the SAP environment. With 12 submitted procedures, the participation was almost twice as high as before. It was shown that the RefMod-Mine / NHCM (further development of the RefMod-Mine / NSCM) and OPBOT (University of Leipzig) have a particularly good evaluation value. Overall, the NHCM algorithm achieved the best results across all domains and was able to beat the competition once again. This enabled us to defend the leading international position within the framework of Business Process Model Matching.
- Contest-Webseite: https://ai.wu.ac.at/emisa2015/contest.php
- RefMod-Miner as a Service: http://rmm.dfki.de (RefMod-Mine/NHCM als freier Service im Web)
- Thaler, T., et al., Evaluating the Evaluation of Process Matching Techniques, in Multikonferenz Wirtschaftsinformatik 2014 (MKWI), D. Kundisch, L. Suhl, and L. Beckmann, Editors. 2014, GITO-Verlag: Paderborn. p. 1600-1612.
- Cayoglu, U., et al., The Process Model Matching Contest 2013, in 4th International Workshop on Process Model Collections: Management and Reuse (PMC-MR’13) 2013.
- Antunes, G., et al., The Process Model Matching Contest 2015, in Enterprise Modelling and Information Systems Architectures, tba, Editor 2015, Gesellschaft für Informatik: Bonn. p. 127-155.