An enhanced automated approach for transforming natural language process descriptions to BPMN2.0 process diagrams – with an evaluation of the application to ISO-Norm process descriptions

This is the code of the bachelor thesis of Vincent Derek Held, written at the chair of Information Systems and Business Process Management at the Technical University of Munich (TUM). The Thesis was written under the supervision of Catherine Sai and under Prof. Dr. Stefanie Rinderle-Ma and can be found in this repository.

Since this work is based on the approach of Shuaiwei YU, in this repository is also published code from him.

Abstract

A comprehensive understanding of business processes is crucial for the digitization of these processes. The utilization of Business Process Model and Notation (BPMN) 2.0 process diagrams has emerged as a pivotal tool in both research and industry for representing and analyzing business workflows. These processes are influenced by an ever growing amount of regulatory documents and process execution data.

This thesis is a contribution to develop a state-of-the-art approach for transforming natural language process descriptions to BPMN2.0 process diagrams. The aim is to evaluate how recent developments have evolved compared to existing methods and how well the approach works for process descriptions from more complex regulatory documents (e.g. ISO standards or data protection regulations). Furthermore, it will be investigated which technologies are best suited to extract information and how to visualize the process in a BPMN2.0 process model.

Keywords: Natural Language Processing, Business Process Compliance, Natural Language to Process, Business Process Model Generation

Installation:

1. Download the gitHub repository
2. Install the required packages with the right version:

pip install -r requirements.txt

3. Activate the following line in the "start_task" Method in the "BPMNStarter.py" to download all dependencies the first time

download_all_dependencies()

Execution:

1. Enter the input path to the textual description (.txt), to use it as an input file
2. Enter the output path, where the generated diagram should be stored (.png)
3. Execute the "main" methode in the "main.py" file

FAQs or Common Issues

1. tbd.
2. tbd.
3. tbd.