Towards Scalable Process Mining Pipelines

Over the past two decades, process mining has proven to be a valuable approach to gain insights into organizations' performance. The major sub-fields of discovery, conformance, and improvement have witnessed substantial development. Contributions have covered the spectrum of better algorithms, richer comparison metrics, and movement towards online analysis for process data. Mostly, these contributions were addressing process mining guidelines from the process mining manifesto. In this paper, we address the sixth guideline in the process mining manifesto. That is, process mining should be a continuous process. For this, we propose a pipelining approach that is: configurable, scalable, modular, and automated. We realize our proposal using Dask and evaluate it with different architectures, process discovery, and evaluation metrics.