About the separate configuration

[HalilB84]

Hi anvaka, big fan of your work.

I am building my version of an interactive map of Wikipedia, and I was curious about this statement in the README:

"In the fourth phase, I used my own ngraph.forcelayout to compute layouts of nodes inside clusters, and a separate configuration to get the global layout of clusters."

Could you explain the last part in more detail? Did you apply another force-directed layout to the 1000+ islands (essentially treating each island as a single node)?

Because the Wikipedia dataset is huge, I am working with 500k–6m nodes. Like in your projects, I prefer the communities to have distinct islands that do not overlap. Right now, my custom (and very naive) graph layout is fast, but it is not accurate—similar communities are very far apart. I would like to achieve a force-directed layout look, but with 6m nodes, performance becomes an issue.

As a solution, I had a similar idea: dividing every article into a large number of communities using the Leiden algorithm and then applying a force-directed layout to all the islands for final placement. The implementation is still a bit unclear in my mind, and I’m not sure how it will work, but that is my current approach. If you could provide insights into how you achieved the island look, I would greatly appreciate it.

Also, your Map of Reddit project has been the main inspiration for my work—thanks for the amazing work!