Time-Order Map for Seamless Zooming between Process Models and Process Instances

Process mining visualizations provide analysts with relevant cues for identifying bottlenecks, waiting times, or other operational inefficiencies. Such visualizations show instances of the process and the corresponding model separately. This spatial separation requires users to mentally integrate detailed and generalized data representations, hindering effective traceability between the two perspectives. This study introduces an interactive visual analytics technique for process mining that seamlessly links the process model view with the corresponding case view through a zoom-based visualization. We first define a two-dimensional coordinate system for plotting individual event sequences. Then, we define semantic layers using contour diagrams to incrementally aggregate nodes and edges into a directly-follows graph. Finally, the resulting Time-Order Map enables non-geometric semantic zooming, allowing for a seamless transition between multiple abstraction levels of a process graph, including its instance representation. We evaluate our technique using realworld datasets to demonstrate its effectiveness in exploring the connection between the process model and the underlying cases. In this way, our approach offers a solution for semantic zooming in process mining, facilitating navigation between process view levels.