.png){kind=icon}
Return
Towards Trace Variant Explainability
.png){kind=icon}
PRE
AI
DOI：10.1007/978-3-032-05281-0_7.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Understanding process variability remains a key challenge in process mining, especially when analyzing trace variants that capture diverse behavioral patterns in modern systems characterized by high data volume and complexity. In this paper, we propose a multi-view explanation framework that enables interpretable analysis of trace variants through both global and local explanations. Our approach first encodes traces using multiple views-including activity, transition, case, and event-level information-applying nominal and time-based strategies suited to structural and performance aspects. Then, it introduces a dual-layer explanation mechanism: at the global level, variant representatives are compared to a global behavioral profile to highlight key differentiating features; at the local level, individual traces are contrasted with their assigned representatives to reveal the specific factors driving their inclusion. The framework is compatible with various variant analysis techniques, as long as a representative (e.g., medoid or central node) can be defined. We demonstrate the method's effectiveness using benchmark event logs and interpretable visualizations, showing how it supports transparency, trust, and diagnostic insight in trace variant analysis.
Keywords:
Process mining
Conformance Checking
Clustering
Explainability
Variant analysis
Multi-perspective
Journal
A
IF:
0
Papers:
19
Citations:
0
Organization
Cited Papers
ShareSaveA Lightweight BPMN Extension for Business Process-Oriented Requirements Engineering.png)
Computers
ComputersIF0
PREAI
ShareSaveExplainable Artificial Intelligence (XAI) 2.0: A manifesto of open challenges and interdisciplinary research directionsINFORMATION FUSION
OAAI
INFORMATION FUSIONIF15.5
OAAIShareSave
ShareSaveShareSaveShareSave
View more