New Region-Based Algorithms for Deriving Bounded Petri Nets

Abstract

The theory of regions was introduced in the early nineties as a method to bridge state and event-based models. This paper tackles the problem of deriving a Petri net from a state-based model, using the theory of regions. Some of the restrictions required in the traditional approach are dropped in this paper, together with significant extensions that make the approach applicable in new scenarios. One of these scenarios is Process Mining, where accepting (discovering) additional behavior in the synthesized Petri net is sometimes valued. The algorithmic emphasis used in this paper contributes to the demystification of the theory of regions as been only a good theoretical exercise, opening the door for its application in the industrial domain.

An efficient process mining model for Petri Nets in process discovery

“…It orders events sequentially, such that each event refers to a case and activity. It has problems with noise, infrequent behavior and complex routing constructs [14]. It scans event logs for particular patterns.…”

“…The main objective of this paper is to propose a conversion for k-bounded discovery that produces more compact TSs than the traditional conversion methods, without losing any precision [22]. Although this technique can be applied to any TS regardless of the conversion used, here develop scheme on top of the sequence and multiset changes, since the other changes are not guaranteed to reserve the same set of regions as the other two do [23]. For region-based discovery algorithms yielding a k bounded PN, this state equivalence description can be distinguished, since the algorithm will only reflect the k bounded regions of the TS.…”

Process Mining Meets Abstract Interpretation

“…Besides the work related to the theory of regions cited above [4,20,5], there are other approaches for process mining. In [19], an algorithm (called α-algorithm) to derive a restricted class of Petri nets was presented.…”

“…More concretely, the goal is to obtain a PN whose behavior contains all the traces in the log, but maybe more. Within this area, several algorithms have been proposed to accomplish this task [4,20,5], most of them based on the theory of regions [10]. Informally, the theory of regions tries to map structures in the state-based or language-based representation of a system into places of the derived PN.…”

Process Model Discovery: A Method Based on Transition System Decomposition

“…Later this approach was generalized by J. Cortadella et al [12,13]. An alternative generalization was proposed by J. Carmona et al [11]. The application of statebased region algorithms to process mining was studied in [6,9,21].…”