This repository provides a baseline framework for extracting Ontology Design Patterns (ODPs) in the Materials Science and Engineering (MSE) domain.
It accompanies the paper:
Semantic Representation of Processes with Ontology Design Patterns
[Norouzi, Ebrahim, Sven Hertling, Jörg Waitelonis, and Harald Sack]
arXiv:2509.23776 — https://arxiv.org/abs/2509.23776
- GroundtruthTerms/ – Contains manually curated ground truth terms for evaluation.
- ODPs/ – ODPs useful for the MSE domain.
- Ontologies/ – Source ontologies used for pattern extraction and evaluation.
- Patterns/ – Extracted and manually curated ontology design patterns per requirement.
- ODP_Extractor_baseline.ipynb – Jupyter notebook implementing the baseline method for semantic similarity-based pattern extraction.
- Latex.ipynb – Notebook to support LaTeX output generation for documentation.
- run.bat – Batch script for executing the ROBOT tool for the ontology modules construction.
The extraction pipeline uses semantic similarity matching between textual requirements and ontology elements (IRIs), transforming textual descriptions into embedding-based queries. The approach supports identifying reusable patterns across different ontologies such as PMDcore, P-PLAN, M4I, and GPO.
- Prepare requirements and the Ground Truth dataset in
Miro Board. - Run ODP Extractor baseline to extract matching ontology elements.
- Review and organize resulting IRIs into candidate ODPs in
ODPs/.
The evaluation framework assesses precision, recall, and F1-score per pattern category:
- Process ODP (Requirement 1)
- Resource ODP (Requirement 2)
- Project ODP (Requirement 3)
Performance varies depending on annotation richness and alignment of the ontology with MSE-specific semantics.
If you use this repository or methodology in your research, please cite:
@article{norouzi2025processODPs,
title={Semantic Representation of Processes with Ontology Design Patterns},
author={Norouzi, Ebrahim, Sven Hertling, Jörg Waitelonis, and Harald Sack},
journal={arXiv preprint arXiv:2509.23776},
year={2025},
url={https://arxiv.org/abs/2509.23776}
}
## License
[MIT License](License)