DisCoveR: Accurate & Efficient Discovery of Declarative Process Models

TL;DR

DisCoveR is a highly efficient and accurate algorithm for discovering declarative process models, specifically DCR Graphs, from event logs, outperforming existing methods in speed and accuracy, and enhancing user modeling experience.

Contribution

The paper introduces DisCoveR, a novel, highly efficient declarative process miner that significantly improves speed and accuracy over existing tools for DCR Graphs discovery.

Findings

DisCoveR achieves 96.2% accuracy in process discovery.

DisCoveR runs 10-100 times faster than previous methods.

Integration of DisCoveR improves end-user modeling experience.

Abstract

Declarative process modeling formalisms - which capture high-level process constraints - have seen growing interest, especially for modeling flexible processes. This paper presents DisCoveR, an extremely efficient and accurate declarative miner for learning Dynamic Condition Response (DCR) Graphs from event logs. We precisely formalize the algorithm, describe a highly efficient bit vector implementation and rigorously evaluate performance against two other declarative miners, representing the state-of-the-art in Declare and DCR Graphs mining. DisCoveR outperforms each of these w.r.t. a binary classification task, achieving an average accuracy of 96.2% in the Process Discovery Contest 2019. Due to its linear time complexity, DisCoveR also achieves run-times 1-2 orders of magnitude below its declarative counterparts. Finally, we show how the miner has been integrated in a state-of-the-art declarative process modeling framework as a model recommendation tool, discuss how discovery can play an integral part of the modeling task and report on how the integration has improved the modeling experience of end-users.