![[2602.16715] Retrieval Augmented (Knowledge Graph), and Large Language Model-Driven Design Structure Matrix (DSM) Generation of Cyber-Physical Systems](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.16715] Retrieval Augmented (Knowledge Graph), and Large Language Model-Driven Design Structure Matrix (DSM) Generation of Cyber-Physical Systems
Summary
This article discusses the use of Large Language Models and Retrieval-Augmented Generation for generating Design Structure Matrices in Cyber-Physical Systems, showcasing their application in two case studies.
Why It Matters
The integration of LLMs and RAG in generating Design Structure Matrices represents a significant advancement in automating the design process of Cyber-Physical Systems. This research could enhance efficiency and accuracy in systems engineering, making it relevant for both academia and industry.
Key Takeaways
- Large Language Models can automate the generation of Design Structure Matrices (DSMs).
- The study evaluates LLMs on two case studies: a power screwdriver and a CubeSat.
- Performance is measured by assessing relationships between components in the DSM.
- The research identifies opportunities for improving automation in systems design.
- All code is publicly available for reproducibility and expert feedback.
Computer Science > Artificial Intelligence arXiv:2602.16715 (cs) [Submitted on 30 Jan 2026] Title:Retrieval Augmented (Knowledge Graph), and Large Language Model-Driven Design Structure Matrix (DSM) Generation of Cyber-Physical Systems Authors:H. Sinan Bank, Daniel R. Herber View a PDF of the paper titled Retrieval Augmented (Knowledge Graph), and Large Language Model-Driven Design Structure Matrix (DSM) Generation of Cyber-Physical Systems, by H. Sinan Bank and Daniel R. Herber View PDF HTML (experimental) Abstract:We explore the potential of Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), and Graph-based RAG (GraphRAG) for generating Design Structure Matrices (DSMs). We test these methods on two distinct use cases -- a power screwdriver and a CubeSat with known architectural references -- evaluating their performance on two key tasks: determining relationships between predefined components, and the more complex challenge of identifying components and their subsequent relationships. We measure the performance by assessing each element of the DSM and overall architecture. Despite design and computational challenges, we identify opportunities for automated DSM generation, with all code publicly available for reproducibility and further feedback from the domain experts. Comments: Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Systems and Control (eess.SY) Cite as: arXiv:2602.16715 [cs.AI] (or arXiv:2602.16715v1 [cs.AI] for ...
