When I use the global graph view, I watch out for clusters os notes: if there’s two or three bigger nodes close together, I zoom in on them and look around. Sometimes, I discover curious (and useful) affinities between subjects or times periods.
For a more focused experience, I use the local graph, with a depth of 2. That gives me a “Wikipedia rabbit hole” kind of flow.
Both of these uses are somewhat serendipitous and irreproduceable. I cannot build a fixed graphic representation of ideas, I just dive into this messy, temporary snapshot of my notes.
When I need mind maps, I still use mind-mapping software.