Header links vs Seperate notes?

Hey everyone!
I understand that this might be a preference thing, but do guys strictly create links to new notes, and keep them small, or do you like to keep some related topics in one file and create links to the headers?

Im quite new to both zettlekasten and roam, so want to get a good start, and looking for pros vs cons.
Also is there any performace difference in roam depending on the approach i take?

For example, atm I’ve created ID’s and links to the headers Org-roam and Org-mode, which are headers in Emacs → Installed packages → Org…

I have been using the Headers UID most of the time, but ever since I have incorporated the concept of “MOC” into my PKM have started to use separate notes. Since org-roam doesn’t have the block-referencing feature, this has also helped me to ignore the ick that I have had ever since I used “Dendron”. There is a package called Dendroam by vicrdguez.

I think if you want you can try to utilise it, it might help u have a clear idea of which notes are the children and which are the parent notes. This way, you can have the functionality of both the worlds.

1 Like

Haven’t heard of that package before, so i will definitely look into it, thanks! :slight_smile:
And yeah i feel like having UID on some headers make things clearer for me, but a part of me also feel like I shouldn’t be inconsistent with the way I’m creating notes

There Is another way which the maintainer of org-roam recommends.

but a part of me also feel like I shouldn’t be inconsistent with the way I’m creating notes

I also have this fear, hence I have decided to use the UDC System I won’t say it’s the best, but I feel confidence in following a system for categorization I ues ChatGPT to help me.

1 Like

Thanks for the link to the right org-roam wiki page :slight_smile:
Wow, i just took a quick look at the UDC System and it seems complicated :sweat_smile:

I might take a closer look at UDC when i have the time, but i would prefer to have the note creation be simple/smooth once i have the inital setup done.

Even I had a headache, so what I did was choose the section of the UDC which I frequently take notes on and added it to my MOC[Alexandria] then I linked the heading to my sub-MOC.

like this

To classify the topic "Language Learning: But Acquiring" in the Universal Decimal Classification (UDC) system, you might consider placing it under the category related to language and education. A suitable classification could be:

  • UDC Hierarchy
    ** 8 Linguistics. Literature
    *** 81 Linguistics and languages
    **** 81-3 Language education
    ***** 81-37 Language learning
    ****** 81-37:8 But acquiring

In this example:

- "8 Linguistics. Literature" is the broader category.
- "81 Linguistics and languages" is the subcategory related to linguistics and languages.
- "81-3 Language education" is a more specific category within linguistics and languages.
- "81-37 Language learning" further refines the focus to language learning.
- "81-37:8 But acquiring" provides additional specificity within the context of language learning.

Please note that the UDC system may have more detailed or alternative subdivisions, and this is a simplified representation. Always refer to the latest UDC schedules and guidelines for the most accurate classification.

So I will create a tag :UDC81-37: and then create the sub-notes for the UDC81 or just link it to the heading with a UID if I think this will be just a very small list.

Also have a look at VULPEA. It might also help you in having a tree like hierarchy.

I think you should keep it simple. I keep things in 1 topic based org file with subheadings that have IDs, so they’re linkable and jump-able via org-roam-node-find, but also I can just go to numpy.org and search the buffer for stuff if I’m looking for numpy things.

Once a topic starts getting too big, I’ll do some refactoring with org-roam-extract-subtree and split it out. For example I used to have just one linux.org with a lot of stuff, but eventually I had enough notes related to “networking” topics that I split then out to a networking.org that’s linked from linux.org.

In my experience, if the notes get too tiny and too numerous, it starts to get harder to grok them all and you really have to lean on the tools to make sense of the graph. While the tools are great, it’s nice to have the option to just read a buffer.

I’d play around with it and see what you like, though, since it feels like mostly personal preference. But I wouldn’t worry too much about consistency or getting “stuck” with a bad method; refactoring is easy!

1 Like

So today I was surfing the web as I was tinkering around with my Emacs init file. I came across this new (it’s new to me) package which will essentially do the work automatically based on the tags and the regex that you configure it to run (need to read it in detail). This is the aforementioned package

Also, look at this discussion of a particular package. I think this will eliminate this confusion of having a separate note or a using header link, as you can at any time use this package to separate the header to a separate note.