On the use of hierarchical subtrace mining for efficient local process model mining

Mining local patterns of process behavior is a vital tool for the analysis of event data that originates from flexible processes, for which it is generally not possible to describe the behavior of the process in a single process model without overgeneralizing the behavior allowed by the process. Several techniques for mining such local patterns have been developed throughout the years, including Local Process Model (LPM) mining and the hierarchical mining of frequent subtraces (i.e., subprocesses). These two techniques can be considered to be orthogonal, i.e., they provide different types of insights on the behavior observed in an event log. As a consequence, it is often useful to apply both techniques to the data. However, both techniques can be computationally intensive, hindering data analysis. In this work, we explore how the output of a subtrace mining approach can be used to mine LPMs more efficiently. We show on a collection of real-life event logs that exploiting the ordering constraints extracted from subtraces lowers the computation time needed for LPM mining compared to state-of-the-art techniques, while at the same time mining higher quality LPMs. Additionally, by mining LPMs from subtraces, we can obtain a more structured and meaningful representation of subprocesses allowing for classic process-flow constructs such as parallel ordering, choices, and loops, besides the precedence relations shown by subtraces.