Mozafari Mehr, A. S. (2024). Multi-perspective Conformance Checking: Identifying and Understanding Patterns of Anomalous Behavior. [Phd Thesis 1 (Research TU/e / Graduation TU/e), Mathematics and Computer Science]. Eindhoven University of Technology.
Summary
The problem of anomaly detection in business process executions has high level of complexity. On one hand, detecting deviating behavior requires considering various aspects of a business process and related constraints, such as the correct sequence of activities, skipped or unexpected activities, resource misrepresentation, access control of underlying data, and compliance with standards, business rules, privacy policies, and regulations. On the other hand, the results of the anomaly detection should be explainable and interpretable, allowing users to understand the detected anomalies. This enhances decision-making and enables domain experts to take appropriate actions based on the identified anomalies.
To overcome these challenges, this research presents a framework for conformance analysis of business process executions, aiming to automatically identify and interpret patterns of deviating behavior from multiple perspectives and explain these deviations in their contexts.
The techniques presented in this dissertation have been evaluated in controlled experiments using synthetic event logs as well as through exploratory experiments with real-life event logs and processes. The results show that our techniques help the user to process and analyze event data and gain insights into system and human behavioural patterns in business process executions. The implementations of these techniques are publicly available as a package in the ProM framework