Tax, Niek, Sidorova, Natalia, van der Aalst, Wil M.P. & Haakma, Reinder (2018). Local process model discovery : bringing petri nets to the pattern mining world. In V. Khohamenko & O.H. Roux (Eds.), Application and Theory of Petri Nets and Concurrency (pp. 374-384). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), No. 10877 LNCS). Dordrecht: Springer Netherlands.

Abstract

This paper introduces the tool LocalProcessModelDiscovery, which is available as a package in the process mining toolkit ProM. LocalProcessModelDiscovery aims to discover local process models, i.e., frequent patterns extracted from event logs, where each frequent pattern is expressed in the form of a Petri net. Local process models can be positioned in-between process discovery and Petri net synthesis on the one hand, and sequential pattern mining on the other hand. Like pattern mining techniques, the LocalProcessModelDiscovery tool focuses on the extraction of a set of frequent patterns, in contrast to Petri net synthesis and process discovery techniques that aim to describe all behavior seen in an event log in the form of a single model. Like Petri net synthesis and process discovery techniques, the models discovered with LocalProcessModelDiscovery can express a diverse set of behavioral constructs. This contrasts sequential pattern mining techniques, which are limited to patterns that describe sequential orderings in the data and are unable to express loops, choices, and concurrency.