Graphics Reference
In-Depth Information
Tip
Use node size to emphasize nodes that are most significant in the
graph. You can use non-linear scales to create several visible levels of
significance, starting with a small number of nodes that are most
important.
Node Labels
Labels are a critical part of virtually all forms of graphs. Geographic maps
are an example of a highly evolved form of visualization that uses
sophisticated systems of labeling. Compare that to a graph, where, unlike
in a map, location of elements cannot be learned over time. Yet, labeling
is notoriously poor in graph tools. Labeling everything in a medium- to
large-sized graph quickly becomes too cluttered. In those cases, use graph
analytics to identify the most important nodes and label those.
You should draw labels over the top of nodes, and, in most cases, you should
give them an outline so that they can draw on top of each other on occasion
without rendering the topmost label unreadable. Try to keep labels short. If
thedatacontainslongnames,addashortenedversionfordisplay,andmake
the full version accessible through selection or hover.
Links
Links are what makes a graph a graph. But what they tell us, they tell us
about nodes. Nodes are the principal subject of a graph visualization, and
so, accordingly, node perception is more important than link perception.
To use an audio analogy, if links begin to interfere with the ability to read
nodes, they should be dialed down to be visually quieter so that the nodes
can be heard. For example, you should draw links behind nodes and not use
fully saturated, attention-grabbing colors. Often, transparency is useful in
displaying links because they will naturally then blend with the background
color, making them less visually noisy. As an added bonus, transparency
allows the overlapping of links to be more easily seen. In most cases, you
should either draw links in neutral colors or interpolate between colors of
the nodes at either end.
