Having used Graph View, I believe there is another graphical approach that could be a powerful complement.
Imagine an infinite canvas where you can lay out and (re)arrange notes, similar to laying out a bunch of index cards on a table or a cork board of cards and clippings you might see in a detective story. The point of the canvas is to “play” with the ideas in your notes as you search for associations and relationships. Some of these relationships might be links of the type we already make; if so, you make the link on the canvas by drawing a connector from one note to another and the link is recorded appropriately in the corresponding files.
Beyond (hyper)crosslinks, each note/file is an object (choose shape, size) on the canvas. It can be moved around, like one might move objects around on a Powerpoint document. Related notes can be put close together. Arrows/icons could be added to indicate meaning. These arrows aren’t links but visual cues. Some notes could be “grouped” and moved around as a unit.
The overall idea is to be able to work with notes as pseudo-physical objects, as one might with notecards.
You could have multiple canvases, with each canvas being an independent drawing/cork board. A canvas can be saved, and when re-opened, the notes are where you left them.
The canvas(es) to which a note belongs, its (X,Y) position, and so on could be stored in the front matter (YAML). Alternatively (and perhaps more powerful), a canvas-database file could reside in the root directory with note IDs and canvases to which these notes belong. The latter could also contain information about objects that are not notes (e.g., block arrows), which notes belong in groups, and so on.
EDIT: 1. see Tinderbox and Curio for examples. 2. what I am describing is not a mind map.
Edit: corrected Curio link.