Hacker News new | ask | show | jobs
by j2kun 3840 days ago
Roughly what size networks can you actually get useful information for? Every time I play with Gephi using any network of reasonable size (in my field), it's always a giant useless hairball.
3 comments

I have tested a few layout algorithms in Gephi for networks with ~200K nodes. It becomes slow but does work.
50-100k nodes, as posters above said, it's very variable on the network structure, but for me I get highly information clusters. Try increasing your cutoff for interaction via edge weight might help.
That's due to the network structure rather than the size. Maybe you can strip/split it to sub-'questions' or weigh nodes or edges?