Semantic fusion of SysML elements for model integration utilizing knowledge graphs

ABSTRACT: While top-down System Engineering supports the definition of system boundaries and interfaces, practical implementation often proceeds in a bottom-up manner, resulting in the need for model integration of SysML models. This is currently hindered by inconsistent naming, different abstraction levels, and unclear interfaces. In this work, a novel approach to integration is proposed, utilizing RDF knowledge graphs and LLM-driven entity alignment with similarity thresholds to perform semantic fusion. A practical use case shows correct consolidation based on cosine similarity thresholds.