Using the Power of Graphs to Understand and Navigate Systems of Systems
Authors: Liana Kiff, Sr. Consultant, Tom Sawyer Software, Janet Six, PhD, Tom Sawyer Software
Physical digital systems utilize digital twins for comprehensive lifecycle management and real-time optimization via temporal and/or spatial data. From sprawling industrial infrastructures to global social networks, the challenge is not just collecting data but making sense of the relationships and contexts that define how these systems work. Graph-based data structures and visualization have emerged as essential tools for managing this complexity, especially in the realm of systems of systems—large-scale, multi-layered networks where each component system interacts with others in often unpredictable ways. [1]
Why Context Is Key in Complex Systems
Context is what transforms raw information into actionable knowledge. Every element of data can have meaningful applications within multiple, overlapping contexts. In a system of systems, managing these interwoven contexts is critical—the integrity, reliability, and usefulness of the data depend on it. [2]
Consider a smart city, where transportation, energy, water, and emergency services are all interconnected. Each system generates its own data, but real value emerges when you understand the relationships—how a power outage impacts traffic lights, or how emergency services reroute in response to a road closure. Graph-based structures are uniquely suited to modeling these complex, dynamic relationships.
The Human Factor: Perception and Cognitive Limits
While technology provides the tools to store and connect vast amounts of information, humans remain the ultimate consumers and decision-makers. Research shows clear limits to how much complexity people can handle when visualizing graphs—especially as the number of nodes and connections grows. [3] Many existing studies rely on small, homogeneous groups (often college students), meaning that best practices for real-world use cases and diverse users are still evolving.
With the explosion of sensors and data streams, orienting oneself within a massive information space is increasingly challenging. Users need clear starting points ("You are here") and intuitive tools for navigation, or they risk getting lost in a dense, confusing data forest. Good system design should enhance a user's situational awareness and support effective orientation and decision-making, even under pressure.
Complexity Factors: Beyond Volume
Complexity in systems of systems isn't just about the sheer amount of data. Other factors, such as time, geography, and the layering of different concerns, add further dimensions. Each facet may require its own abstraction or visualization technique to make the data comprehensible to users. Often, generic methods of navigation in complex systems fall short of helping users navigate domain-specific knowledge spaces.
Graph Navigation Strategies: Top-Down, Bottom-Up, and Middle-Out
Navigating complex graphs can follow several strategies, each suited to different user needs [4]:
- Top-Down: Start with an overview of the entire system, then drill down to specifics. This is ideal for monitoring or data science tasks, where understanding the big picture is crucial before focusing on details.
- Bottom-Up: Begin at a specific point of interest and explore outward. Useful for investigations or troubleshooting, such as tracing the source of a system failure.
- Middle-Out: Start from an abstraction or cluster within the graph, then move to more detailed or broader views. This is common when browsing large systems where users may not know their endpoint in advance, such as navigating a wiki or exploring a knowledge graph.
Effective navigation and value extraction further relies on two main categories of user interaction methods:
- Data Actions: Selecting, filtering, or calculating on the data itself.
- Presentation Actions: Adjusting layout styles or visual parameters to clarify relationships. For example, adding annotations to nodes, edges, groups of nodes or edges, or entire projects supports systems engineers in understanding the system as it evolves and in the communication of key findings to other stakeholders.
Combining these approaches helps users manage complexity and maintain orientation as they explore vast data spaces.
Use Case: Systems of Systems in Action
Model-Based Systems Engineering (MBSE) relies on diagrams for both input and communication. The recently approved Systems Modeling Language (SysML) version 2.0 specification [5] uses graph-based modeling, which is scalable and robust to collaborative engineering processes. Graph-based structures allow for automated consistency checks, easier updates, and flexible visualizations—reducing errors over traditional methods. [6]
Automated system navigation is valuable for both design-time engineering and real-time operations. Consider the example of an industrial air handling unit (AHU) within a larger building management system—a classic system of systems scenario.
- Component Complexity: The AHU itself is comprised of sensors, valves, heating and cooling coils, and control logic.
- Interconnected Contexts: The AHU interacts directly with other comfort systems, including the Hot Water Plant and Chilled Water Plant, typically represented as a set of connected graphs.
- Real-Time Monitoring: By layering real-time data onto the graph, operators can quickly see the status of each component and how issues propagate through the system. This supports faster diagnosis and more effective decision-making.
- Scalability: Graph-based approaches scale naturally, allowing for the automated generation of both high-level overviews and detailed drilldowns as needed, without the requirement for manually engineered system-specific displays.
Conclusion: Empowering Human Understanding Through Graphs
Graph-based visualization is more than a technical solution—it is a bridge between overwhelming complexity and human insight. By focusing on context, navigation, and the realities of human cognition, we can design systems that empower users to make informed decisions even in the most intricate systems of systems. The ability to visualize, navigate, and understand interconnected contexts, with the aid of automated graph-based visualization, will be the key to unlocking new levels of efficiency, safety, and innovation.
This blog is based on the article, "Navigating Contextual Complexity with Graph Visualizations" as published in the Journal of Innovation, February 2025 [1], which provides a more comprehensive discussion of the existing research related to complexity and cognition, with further references.
Works Cited
[1] L. M. Kiff and J. Six, "Navigating Contextual Complexity with Graph Visualization," Journal of Innovation, no. 26, pp. https://www.objectmanagementgroup.org/wp-content/uploads/sites/8/2025/02/JOI_20250204_5_Navigating_Contextual_Complexity_with_Graph_Visualization_Standalone.pdf, February 2025.
[2] H. Li, G. Appleby, C. D. Brumar, R. Chang and A. Suh, "Knowledge Graphs in Practice: Characterizing their Users, Challenges, and Visualization Opportunities," vol. 30, no. 1, pp. 584-594, Jan 2024.
[3] V. Yoghourdjian, D. Archambault, S. Diehl, T. Dwyer, K. Klein, H. Purchase and H.-Y. Wu, "Exploring the Limits of Complexity: A survey of empirical studies on graph visualization," Visual Informatics, pp. 264-282, 2018.
[4] T. von Landesberger, A. Kuijper, T. Schreck, J. Kohlhammer, J. van Wijk, J. Fekete and D. Fellner, "Visual Analysis of Large Graphs: State-of-the-Art and Future Research Challenges," Computer Graphics Forum Volume 30, Issue 6, pp. 1719-1749, 2011.
[5] Object Management Group, "About the OMG System Modeling Language Specification Version 2.0 beta 4," 22 July 2025. [Online]. Available: https://www.omg.org/spec/SysML. [Accessed 22 July 2025].
[6] M. Bajaj, J. Backhaus, T. Walden, M. Waikar, D. Zwemer and C. Schreiber, "Graph-Based Digital Blueprint for Model Based Engineering of Complex Systems," in Annual INCOSE International Symposium (IS 2017), Adelaide, Australia, 2017.