Detecting cross-case associations in an event log: toward a pattern-based detection

Yael Dubinsky, Pnina Soffer, Irit Hadar

Research output: Contribution to journalArticlepeer-review


Business process management, design, and analysis is mostly centered around a process model, which depicts the behavior of a process case (instance). As a result, behavior that associates several cases together has received less attention. Yet, it is important to understand and track associations among cases, as they bear substantial consequences for compliance with regulations, root cause analysis of performance issues, exception handling, and prediction. This paper presents a framework of cross-case association patterns, categorized as intended association patterns and contextual association patterns. It further conceptualizes two example patterns—one for each category, and proposes techniques for detecting these patterns in an event log. The “split-case” workaround is an example of a pattern in the intended association category, and its proposed detection method exemplifies how patterns in this category can be approached. The patterns of a shared entity and a shared resource are contextual association patterns, which we propose to detect by means of hidden concept drifts. Evaluation of the two detection approaches is reported, using simulated logs for assessing their internal validity as well as real-life ones for exploring their external validity.

Original languageEnglish
JournalSoftware and Systems Modeling
StateAccepted/In press - 2023

Bibliographical note

Funding Information:
The research was supported by the Israel Science Foundation under grant agreement 669/17.

Publisher Copyright:
© 2023, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.


  • Cross-case patterns
  • Process mining
  • Split-case workaround

ASJC Scopus subject areas

  • Software
  • Modeling and Simulation


Dive into the research topics of 'Detecting cross-case associations in an event log: toward a pattern-based detection'. Together they form a unique fingerprint.

Cite this