0

I have a (currently in-progress) directed graph that I am creating with Dot. The attached image is what I have so far. Graph so far As you can see, it is hard to read because it is difficult to tell what labels correspond to which edges. I expect the graph to grow much, much larger. I tweaked Dot's settings to increase nodesep and ranksep, and this helped, but I am curious if there are any other things I can do. Here is some info about the graph:

  • I cannot remove many vertices. If size = convoluted and hard to read, so be it, but I want at least for edge labels to be clear on which edge they are labeling.
  • The vertices can be broadly broken down into ten sections. (The vast majority of those in the graph so far are in section 1, with maybe 5 at most being in sections 2 or 3). Most edges will be between vertices of the same section, so subgraph divisions may help, but some edges will be between sections.
  • I anticipate later sections may be smaller than the first section, but this is not guaranteed.
  • Within a section, edges are arbitrary. They may connect from anything to anything else, you name it. There is no structure that may be relied upon (ie tree-like structure).
  • I expect each section to have one or two "core" vertices with a very high connectivity. The core vertex of section 1 is "Adokoi". There may be utility in hand-selecting core vertices and instructing Dot to place these in a special way.
  • At the moment, the only customizations/special instructions I have done are increasing nodesep and ranksep, and making the graph bottom to top. All other possible changes are on the table.

With this in mind, what options do I have for things like breaking into subgraphs, special instructions for which rank things go into, etc, in order to improve the readability and user friendliness of this graph?

1 Answer 1

0

The larger the graph, the harder it gets to "organize" it...

One might get something more readable, for example,
by using the "record" style for the major nodes (those with most edges attached).
i.e. by that keeping them "together"...
or by similar grouping in other ways.

Ref:
https://graphviz.org/doc/info/shapes.html#record

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .