PM4Py.LLM: a Comprehensive Module for Implementing PM on LLMs

TL;DR

The paper addresses the challenge of applying process mining to LLM-enabled workflows while safeguarding privacy and mitigating hallucinations. It presents pm4py as a comprehensive platform that implements multiple paradigms for PM on LLMs, including textual/visual abstractions, SQL-based querying, code generation, and automatic hypothesis formulation. The authors detail concrete implementations for both traditional event logs and object-centric logs (OCEL), with practical tools for data representation, visualization, and executable code. The work demonstrates how PM-on-LLM can enable flexible querying, hypothesis testing, and autonomous analysis, and discusses future directions such as video-to-log extraction and autonomous PM tasks.

Abstract

pm4py is a process mining library for Python implementing several process mining (PM) artifacts and algorithms. It also offers methods to integrate PM with large language models (LLMs). This paper examines how the current paradigms of PM on LLM are implemented in pm4py, identifying challenges such as privacy, hallucinations, and the context window limit.