alt Hacker NewsIt's a common problem to get excited about networks, build a large one, and then by stuck with an unapproachable hairball. If you want to explore network structure, consider using tools like quadrilateral simmelian backones which can provide an opinionated look at what matters in the network.
Replies
tomstuart • today at 7:07 AM
I had to look this up: https://doi.org/10.7155/jgaa.00370
One could also try to use a different set of definitions better suited to such a visualization.
The Oxford Advanced Learner’s dictionary has an appendix called “Defining Vocabulary”. It says:
“In order to make the dictionary definitions easy to understand, we have written them using only the words in the following list.
[…]
Occasionally it has been necessary to use in a definition a word not in the list. When such a word occurs it is shown in SMALL CAPITAL LETTERS.”
I estimate that list has about 3,500 words.
⇒ If you base your network on that dictionary or one carefully constructed like that, the graph could have a central core of about 3,500 nodes with the other words circling around it.
Making a good visualization still would be a challenge, of course.