This is the code used in the thesis "Improved auto-generation of business process models from natural language texts of various complexity" by shuaiwei Yu. The code used to generate the BPMN diagram is contained in the project folder.
This is the entry point of the system, you can specify the input path of the textual descriptions, the title of the generated diagram and the path to save the generated image. After the generation is completed, a piperflow syntax used to render the BPMN diagram will be generated, for further reading please refer to https://github.com/csgoh/processpiper/wiki.
If you are not satisfied with generated output, you can always modify the generated piperflow syntax by yourself. Once your modification is finished, you can use this file to rerender the BPMN diagram by put your modified results into the input_syntax variable.
You can find the BPMN diagram generated by our system and the diagram modified using GPT in this folder.
You can find the used textual description as input in this folder. The generated piperflow syntax is also stored in this folder.
The used PET data set is stored here, a simple tool of visualization is provided.
If you have the PET json data, you can use this script to generate a markdown file with PET tokens labeled with different colors.