Navigating Contextual Complexity with Graph Visualization
Graph-based data structures provide the flexibility required to encode and analyze the increasingly vast , complex , and connected data spaces we are faced with in modern systems . The use cases for graph visualization appear in many disciplines . This paper will focus on use cases relevant to human understanding and analysis of systems-of-systems models and knowledge graphs .
Context is the source of meaning in a data-centric world . For each data content element there may be many legitimate contexts described by unique meta-data . Managing the coherence of those connected contexts within a system of systems is a critical factor impacting the integrity of every potential use case for each piece of managed data .
Graphs provide another tool for communication , but graphs run into measurable human perception and cognition limitations . This paper discusses the practical applications of advanced graph-based navigation and visualization techniques for aiding human comprehension and interaction with complex , multi-layered systems . We cite specific use cases and examples , with reference to deeper research .
1 THE HUMAN FACTOR “ The more things change , the more they stay the same .” 1
There has been a nearly continuous tug-of-war between the performance and convenience of storing information as opposed to the needs of people to consume and manipulate it . The use of graphs for storage doesn ’ t eliminate this battle . Graph-based databases and visualizations only move the limits of what is possible in certain directions and deliver new ways to express the challenges . The same tensions still apply , the same principles are still at work , and many of the same technical solutions are still employed .
If graphs allow for greater facility to link related information , it is also the case that you can build a graph that makes it hard to deliver the answer to the question your user wants to ask . Furthermore , if the data and the human brain now bear more resemblance to each other in terms of networked knowledge , humans are still functionally limited by human faculties of perception .
In their 2018 paper , Vahan Yoghourdjian and colleagues [ 1 ] review and summarize the findings of the best available studies to date regarding the human cognitive limits encountered when working with graphs of different size ( number of nodes ) and density ( number of edges ). One of their conclusions is that not enough work has been done to adequately understand the best solutions and guide designers to the most appropriate knowledge graph visualization techniques for their purpose . Many studies so far have included limited numbers of users , and the major source of those users ( college students ) are not necessarily representative of the real-world user base . Burch et . al . [ 2 ] concluded in their survey of empirical studies that 20 years of research has
1
Jean-Baptiste Alphonse Karr - 1849 70
February 2025