Work

Addressing Complexity in System of Systems With GraphRAG: An AI‐Driven Framework for Dynamic Data Integration

Year: 2025

Type: article

Abstract: ABSTRACT System of Systems (SoS) environments are inherently complex, involving numerous operationally and managerially independent component systems with hidden interdependencies and frequent interactions based on unstructured data. In this paper, we propose using graphical Retrieval‐Augmented Generation (GraphRAG), a tool that combines large language models with knowledge graph (KG) techniques to address these challenges. Using metrics from information entropy and graph theory, we demonstrate how KG construction and clustering nodes can reduce complexity in SoS. An example application in the Urban Air Mobility setting illustrates that GraphRAG can solve concrete data integration challenges while outperforming traditional Retrieval‐Augmented Generation (RAG) methods. The upshot is improved execution of model‐based systems engineering in an SoS context: mitigating risks from incomplete information, enhancing system integration, and improving decision‐making.

Source: Systems Engineering

DOI: 10.1002/sys.70012

Authors: Yinchien Huang, Tien‐Yueh Fung, Daniel A. DeLaurentis

Institution: Purdue University West Lafayette

Language: English

Topic: Systems Engineering Methodologies and Applications

Subfield: Control and Systems Engineering

Field: Engineering

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: hybrid

Estimated APC paid: $3,140