Differences or similarities in networks are typically discovered in a side-by-side comparison.
We can simplify this process by visualizing the differences or similarities between two hive plots as a differential hive plot.
Differential hive plots can be constructed by performing a set operation on two given hive plots "A" and "B":
|intersection||A ∩ B||Select similar nodes and edges from "A" and "B".|
|relative complement||A ∖ B||Select nodes and edges from "A" that are not found in "B".|
|symmetric difference||A Δ B||Select unique nodes and edges from each plot.|
|union||A ∪ B|
Select all nodes and edges from "A" and "B".
Equivalent to joining the results of intersection and symmetric difference.
Here are 4 different differential hive plots:
Definition of Similarness
Two edges are similar if all of the following are met:
- sources are on the same axis
- sinks are on the same axis
- directionality must agree for directed edges only
- positions of the sources are within the margin of error
- positions of the sinks are within the margin of error
- sources have the same name (optional)
- sinks have the same name (optional)
Margin of Error
The margin of error defines the degree of similarness. There are 2 types of error:
The difference in the original values used to define the node's position on the hive plot.
Domain: same as the parameter chosen for node position assignment
The difference in the node's relative position on the axis.
Attributes for Format Rules
Every node or edge in a differential hive plot is assigned a value for the
|This item is unique to hive plot "A".|
|This item is unique to hive plot "B".|
|This item is common to both hive plots "A" and "B".|
In the 4 examples above, format rules were applied to assign colors based on the
Nodes that are not similar are assigned a value for the
|Nodes are on the same axis but no within the margin of error.|
|Nodes are on different axis.|
|The node is not found in the other hive plot.|