A framework to canonicalize manufacturing service capability models

•We propose an OWL-based ontology design pattern based canonicalization framework.•The framework streamlines the OWL-based semantic mediation process.•It decomposes the mapping task.•The resulting mapping statements are simple to comprehend and maintain.•The canonicalization circumvents OWL limitation and increases reasoning performance.

The capability to share precisely defined information models, which reveal a supplier’s manufacturing service capability (MSC) with anyone who needs it, is key to the creation of more agile supply chains. Today, unfortunately, this capability does not exist. Why? Because most suppliers use proprietary information models to represent and share their MSC information! This limits both the semantic precision in the models, which is needed for interoperability, and the level of agility in the supply chains. The availability of a semantically precise and rich reference MSC ontology could address both of these limitations. Based on our prior research, the development of such an ontology will require a semantic mediation process between the proprietary MSC models and the reference MSC ontology. At the heart of every known, semantic-mediation process is a mapping between a proprietary MSC model and the reference MSC ontology. Such a mapping must deal with the structural and semantic conflicts between the two. In this paper, we propose a new approach, which we call canonicalization to address the structural conflicts. The semantic conflicts are addressed using logical mapping. The canonicalization pre-processes the structural representations of the proprietary models and then aligns them using ontology design patterns which are also used in the reference ontology. This simplifies both the mapping problems themselves and the resulting mapping statements considerably. In the paper, we also demonstrate our approach and its benefits in the context of a description-logic-based semantic mediation using the Ontology Web Language (OWL).