A Framework for Processing Textual Descriptions of Business Processes using a Constrained Language -- Technical Report

TL;DR

This paper introduces BeePath, a framework that enables non-experts to create formal process models from plain text descriptions using a constrained language and large language models for translation.

Contribution

It presents a novel framework combining constrained language patterns and LLMs to facilitate process modeling from natural language descriptions.

Findings

Successful translation of plain text into formal models

Framework supports multiple formal modeling languages

Enhances accessibility of process modeling for non-experts

Abstract

This report explores how (potentially constrained) natural language can be used to enable non-experts to develop process models by simply describing scenarios in plain text. To this end, a framework, called BeePath, is proposed. It allows users to write process descriptions in a constrained pattern-based language, which can then be translated into formal models such as Petri nets and DECLARE. The framework also leverages large language models (LLMs) to help convert unstructured descriptions into this constrained language.