Automatisierte Abbildung semantisch heterogener I4.0-Verwaltungsschalen durch Methoden des Natural Language Processing

Systems in the field of Industrie 4.0 should be able to interact with each other in an interoperable manner. In order for this to be realized automatically, they must be semantically interoperable. For this purpose, the current Industrie 4.0 research approach to the interaction of systems focuses on a semantically homogeneous language space. In this paper, we present a method that extends the current approach to include heterogeneous semantics. Mapping unknown vocabularies to a target ontology enables the interactions of semantically heterogeneous Industrie 4.0 asset administration shells. The mapping is based on methods from the field of Natural Language Processing. For this purpose, language models pre-trained on ISO standards and sentence embeddings are combined. This leads to a promising accuracy in the created evaluation dataset, which contains different semantics for identification and design submodels of the Pumpe 4.0 project.