Data Petri Nets meet Probabilistic Programming (Extended version)

TL;DR

This paper introduces a systematic translation of Data Petri Nets into probabilistic programming models, enabling statistical reasoning and process analysis with existing PP tools, supported by a prototype implementation.

Contribution

It provides the first sound translation of Data Petri Nets into probabilistic programming languages, facilitating process mining and analysis using PP systems.

Findings

Translation is sound with statistical guarantees.

Prototype implementation demonstrates practical feasibility.

Enables new process analysis scenarios with PP tools.

Abstract

Probabilistic programming (PP) is a programming paradigm that allows for writing statistical models like ordinary programs, performing simulations by running those programs, and analyzing and refining their statistical behavior using powerful inference engines. This paper takes a step towards leveraging PP for reasoning about data-aware processes. To this end, we present a systematic translation of Data Petri Nets (DPNs) into a model written in a PP language whose features are supported by most PP systems. We show that our translation is sound and provides statistical guarantees for simulating DPNs. Furthermore, we discuss how PP can be used for process mining tasks and report on a prototype implementation of our translation. We also discuss further analysis scenarios that could be easily approached based on the proposed translation and available PP tools.