PM4Py-UCM turns event logs into requirements engineering models

Process mining has long enabled analysts to extract as-is process models (like Petri nets or BPMN) from event logs. However, these outputs have been underutilized in early requirements engineering (RE). Daniel Amyot's new paper introduces PM4Py-UCM, an open-source extension to the popular PM4Py Python library, that makes Use Case Map (UCM) models from ITU-T's User Requirements Notation a first-class discovery output. This allows mined behavior to feed directly into URN-based modeling, analysis, and management activities—bringing evidence from operational data into the earliest stages of system design.

The tool provides a full UCM discovery pipeline, plus novel hierarchical decomposition strategies that produce nested UCM models from flat event logs. It also offers configurable performer mappings for both UCM and BPMN visualizations, and exports to the jUCMNav tool while preserving the model structure under round-trip. Using public and synthetic logs, the paper demonstrates how different abstraction levels and performer groupings render the same behavior differently. The result is a practical instrument for model-driven RE, enabling requirements engineers to ground their decisions in actual system traces rather than assumptions.