Sorry for the long post, I don’t know that brevity is possible:)
Soon after properties was implemented, I made a commitment to cleaning up my vault. It held a lot of carnage from previous attempts to “structure my knowledge” by using complex workflows and following methodologies and related pedagogies - each requires different and complicated sets of metadata and enforced workflows (I suck at the enforced thing). Success in any of those also requires a certain mindset, abilities, and plugins.
What I began to notice was a core set of tags and link names that I used for most of my thinking and writing projects. In the case of nested tags, there were implied hierarchies, many of them overlapped and redundant. Some tags were in yaml but most were peppered thru out notes.
I did try to use graph to assist but I found myself spending more time tweaking the graph criteria than optimizing my vault; but the effort did provide some nice looking and informative graphs which I bookmarked and intend to return to later 
It became obvious at this point that I needed to make a distinction between tags (properties in general) and links.
So I made a rule (unusual for me) - tags and properties ONLY go in yaml frontmatter and they apply to the note.
If I need to distinguish some chunk of a note and don’t want to move that chunk to its own note (atomic, if you will), I make a link out of keywords or phrases in that chunk. Links, to me, make sense in the body of a note. Links are how I get to MOCs (or whatever word you want to use).
Today, for example, I was reviewing an article and found a word called Autotelic and thought I had never heard it before. A quick search revealed I had seen it, added my thoughts, and made links to it but never created the note “Autotelic”. So I did, and, poof! , down in that backlinks section (Linked mentions) at the bottom of each page popped a complete history of the links of that word in my vault and below that was all the unlinked mentions of every occurrence of the word itself; the out of the box functionality in Obsidian is extraordinary!
I have now inserted a yaml frontmatter block at the top of each note with at least one tag at least one property. New web content comes into my vault via MarkDownload and created a template in that browser extension.
I also removed all the nested tags as they became too numerous and confusing, despite the hierarchy. Stay with me…
For all notes I imply a workflow by doing the following:
- a processed_date property - when I bring in notes and don’t get to them right way, I can search for notes that don’t have the property with -[processed_date:]
- a list property called pertaining - if say, the tag for a note is writing and I’m using this note to write about subtext, I’ll assign pertaining to “subtext”.
- a list property called assigned - if this note has a direct relationship to a section in an essay or chapter in a book, I’ll add a link to assigned to the heading in that section or chapter.
- a list property called role - if something in this note applies to a character in a story, I’ll place a link to that role in my character list for the missive.
- each of the above also has a related date property (i.e. pertained_date, assigned_date, and role_date).
- which leads me to another property called status, it is either pertained, assigned, or roleassigned.
For me, this process is iterative. During research for projects, I add significant new properties (including tags) that apply to the new project or have discovered other linkages. I can add more tags and pertaining properties.
I now create hierarchies by a combination of tags and list properties as defined by search query. This is the only way I know how to solve many to many relationships. I bookmark queries and name them accordingly.