This is a question from 5 years ago that I wanted to re-ask.
There are a bunch of graph/network visualization packages out there but often require local binary installation and are kind of difficult for many Jupyter notebook users to get going. We’ve search for a while for something useable in notebooks but haven’t come across anything (and already use Bokeh for several other visualizations).
Most of what you would need to render/visualize/interact with a network is already in Bokeh and the rendering perf would allow it work pretty well with even large graphs.
The things that would be top of my priority list are:
- a decently-sized gallery example of creating a graph with a few hundred nodes that has things like tooltips populated from nx attributes
- dynamic layout and ability to control some parameters interactively (examples I’ve seen use a static layout from an NX layout but not having the glyph layout recalculate dynamically). Ability to add layout algos would be a bonus.
I’d be happy to help with the gallery example but my knowledge of JS and the Bokeh programming model is close to zero.