Dynamic Prompt-Based Framework for Pragmatic Quality Improvement in Business Process Models
Business Process Modeling plays a critical role in enhancing organizational efficiency, standardization, and transparency. While prior research has extensively addressed syntactic and semantic quality dimensions, pragmatic quality, relevant to the clarity, interpretability, and comprehensibility of...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11037667/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Business Process Modeling plays a critical role in enhancing organizational efficiency, standardization, and transparency. While prior research has extensively addressed syntactic and semantic quality dimensions, pragmatic quality, relevant to the clarity, interpretability, and comprehensibility of process elements, remains relatively underexplored. This study presents a dynamic prompt-based framework that leverages Large Language Models to automatically generate meaningful and contextually appropriate labels for BPMN elements, including tasks, gateways, and XOR split outgoing edges. The framework integrates contextual information extracted from BPMN models to construct structured prompts, ensuring that the generated labels are semantically precise and compliant with BPMN labeling conventions. The generated labels are evaluated using the all-MiniLM-L6-v2 sentence transformer model to assess semantic coherence and mutual exclusivity, while spaCy’s syntactic analysis validates verb presence in task labels. Experimental results on the BPMAI dataset, encompassing 29,810 BPMN models, confirm the framework’s effectiveness in enhancing pragmatic quality and improving overall model interpretability. To promote reproducibility and encourage future research, the full implementation and evaluation pipeline are publicly available at <uri>https://github.com/SarahayadAOU/MutuallyExclusive</uri>, and the dataset is accessible via Hugging Face at <uri>https://huggingface.co/datasets/AyadSarah/MutuallyExclusive</uri> |
|---|---|
| ISSN: | 2169-3536 |