Graphics Reference
In-Depth Information
Whereas scaled lines portray a fuzzy impression of magnitude, quantized
line choices enable the viewer to know immediately whether a link value
falls above or below meaningful thresholds.
Tip
In any graph where flow is being represented and expression of link
weights is constrained to a few pixels, quantize line widths to represent
meaningful ranges that the viewer can quickly decode.
Relational Nodes
In most cases, a relation between two nodes (where nodes are the subject
of interest) is expressed as a single link. In certain cases, however, it can be
useful to break out relation types into their own nodes.
For example, if your goal is to identify communities in a social network, a
view of three people who are related as real-life neighbors can benefit from
having links loop through an address relation node of “20-27 Harper Lane,”
or a view of three nodes that share a phone number might be improved by
looping through a common node, “1-800-899-2121.”
Breaking down shared relations into relational nodes enables a single label
to express the relation and helps nodes that share that relation to be drawn
together in layout. This approach is often referred to as a
shared attribute
graph
.
Figure 16-9
shows an example.
When nodes are used to express relations, it is important to visually express
them differently from subject nodes. Relational nodes should be
perceptually recessive relative to primary nodes and should blend
seamlessly with the links they are articulating. You can think of relational
nodes as stepping stones in links, and they should be expressed accordingly.
Tip
Use relational nodes to break out and express common relations when
there is a wide variety of relationship types, and the goal is to
understand communities in a social network. Treat them visually as a
second order of nodes using a palette consistent with the links they are
associated with.
