The goal of a graph is representing the structure, not the appearance of a network. The conversion of a real network into a planar graph is a straightforward process which follows some basic rules:
- The most important rule is that every terminal and intersection point become a node.
- Each connected node is then linked by a straight segment.
The outcome of this abstraction, as portrayed on the above figure, is the actual structure of the network. The real network, depending on its complexity, may be confusing in terms of revealing its connectivity (what is linked to what). A graph representation reveals the connectivity of a network in the best possible way. Other rules can also be applied, depending on the circumstances:
- A node that is not a terminal or an intersection point can be added to the graph if along that segment an attribute is changing. For instance, it would be recommended to represent as a node the shift from 2 lanes to 4 lanes along a continuous road segment, even if that shift does not occur at an intersection or terminal point.
- A “dummy node” can be added for esthetical purposes, especially when it is required that the shape of the graph representation remains comparable to the real network.
- Although the relative location of each node can remain similar to their real-world counterpart (as in the above figure), this is not required. It is however a preferred option to help understand the geographical context of the network being represented.