Interactive Discovery and Visualization of Process Trees

Keywords

Visual Analytics, Process Mining, Process Trees.

Abstract

Process Discovery, a key area of Process Mining, constructs models from event logs containing sequences of activity executions. Current automatic approaches suffer from opaque algorithms, limited incorporation of domain knowledge, and static, hard-to-explore visualizations. To address these limitations, we propose an interactive discovery process involving users in constructing process trees. This allows real-time interaction, dynamic domain expertise integration, and better understanding of the relationship between logs and models. We demonstrate this with a prototype featuring four linked views, and introduce improved process tree visualizations.