What is it about?
Trace Clustering is about grouping together instances of business processes, based on their execution event data. In this work, we aim at making elements of the trace-clustering process more accessible to the decision-makers and enhancing the understandability of the analysis.
Featured Image
Why is it important?
In business processes that are either complex or are executed flexibly, many variants of the same process are possible. This flexibility is observed in the variability of the flows, and it is holding up the potentials of automated process discovery since it leads to “spaghetti” process models. A solution is clustering, but Decision-makers should be more involved in the process. We provide this opportunity by allowing them to guide the clustering process by reinforced or counter-veto effects and pairwise constraints, as well as by proposing a way to trim outliers.
Perspectives
The first situation that turns the proposed approach suitable for trace clustering is the need to take multiple criteria into consideration, especially when those criteria are measured by strongly heterogeneous scales (e.g., ordinal and numerical criteria) while avoiding the need for normalization and for recoding of the evaluation scales. Then, this work allows process analysts to avoid the undesired compensation of a loss on a given criterion by a gain on another one. Last, this work enables DMs to match at a greater extent their preferences to the solution, a situation that is welcome when the subjectivity of DMs is a desired element.
Pavlos Delias
Eastern Macedonia and Thrace Institute of Technology
Read the Original
This page is a summary of: A non‐compensatory approach for trace clustering, International Transactions in Operational Research, February 2017, Wiley,
DOI: 10.1111/itor.12395.
You can read the full text:
Contributors
The following have contributed to this page
