This paper examines in close relation two fields of growing importance: model-based systems engi-neering (MBSE) and natural language processing (NLP). System models provide a structured de-scription of engineering data, whose inherent semantics often remains hard to explore. Natural lan-guage understanding, an important field of NLP, focuses on semantic text comprehension but cannot directly account for structured information sources.
In this paper, we investigate natural language understanding of MBSE artifacts as they appear in industrial scenarios. In this context, the wide and heterogeneous knowledge space and user base calls for novel techniques to facilitate information retrieval from complex system models.
We thus propose to leverage domain-specific text generators to transform models into a descriptive text corpus on which we apply state-of-the-art semantic search and analysis techniques.
We illustrate the approach on relevant MBSE examples by performing a qualitative evaluation of intuitive text search and comprehension in textual descriptions obtained from system models.