# Zettelkasten for scientific research

edited May 2020

Hi everyone!

I'm a computer science student, struggling with note-taking (read: I'm not taking any notes). I'm trying this Zottelkasten graph-based method to improve and motivate myself.

The main reason why I usually don't take notes is that I have a pile of wonderful, painfully written, notebooks in my shelf... that is completely not usable in practice.

If I'm not able to search and retrieve them... they may as well not exist. A forgotten note in a notebook does not exist for me.

The Zottelkasten method seems to solve this problem, so I'm quite excited to try it!

Now the main problem.
I have read a lot of discussions about the Zettelkasten method and every discussion is concentrated on humanistic writing. I want to apply this method to my scientific field, i.e. inserting in the zottelkasten every prior concept needed to generate new ideas.

For example:

• I'm attending a class in Probability: I'd like to insert the topics and notes discussed in that class into the Zottelkasten.
• I read a paper: I'd like a summary for the paper that uses some ideas (links to them).
• I have an idea? I'd like to link that note to the prior concepts needed to understand that idea.
• I learn a new concept? Insert into the db.

How would you approach this problem without making the Zottelkasten unmanageable?
I fear that the mix of different type of notes (class notes, my ideas, paper I read will create confusion in the long run)

I'm thing about a possible approach: basic folder division and an (evolving) tag system.

I was thinking to name each file only with their ID and divide the notes in these folders:

• fleeting: temporary note, before I forget what I wanted to write.
• literature: summary or note that directly comes from some scientific paper or other sources of scientific information. The idea is to avoid to re-read the same stuff twice. This note would decouple and link all the main ideas in the paper into other existing or new notes.
• knowledge: notes that explain every prior concept needed (e.g. a possible note: "What is the likelihood?"). While attending a class I would put here the notes of every topic explained in that class, decouple into atomic ideas and linked togheter.
• permanent: notes from my original ideas or a concentration of literature and knowledge notes that I consider particularly important and should never be forgot. They would be much more refined
• project: notes relative to projects

And a tag system to ease the search and retrieval of notes in several ways:

Meta notes:

- #meta: meta notes about the rules in this knowledge base itself.
- #zettelkasten: related to the Zettelkasten method in some way.

Structure notes:

- #connection:  specific notes whose purpose is to link together other notes and explain their relationship.
- #outline:  notes that contain a sequence of links to other notes in a particular order to create a story, narrative, or argument.
- #set: notes that contain a set of links to other notes to create a cohesive view. Similar to tags, with the possibility to motivate the links.

Note type:

- #idea: an idea that may be further developed into something.
- #original: idea or note that I came up principally on my own.
- #exercise: it is an exercise and relative solution
- #definition: it is purely a definition, nothing fancy.

Note context:

- #paper: it is the summary of a paper I read.
- #course: it is the outline or summary of a course. Usually only the outline is tagged.
- #talk: it is the outline or summary of talk.

Note projects:

- #projectX: related to the projectX
- ...

Topics:

- #learning: related to deep learning and machine learning in general.
- #nlp: related to Natural Language Processing.
- #graph: related to graphs in some way.
- #probability: related to probability concepts in some way.
- #manifold: related to the manifold in some way.
- #information-theory: related to the information theory in some way.
- #redundancy: related to methods to introduce redundant information.
- #error-correction: related to methods that correct errors in binary messages.
- #error-detection: related to methods that detect error in binary messages.
- #shannon: related to Shannon in some way.
- ...


Every note will more or less follow this template with an ID YYYYMMDDhhmmss for the filename:

---
title: 'Note template'
---

# Note template

A sample note showcase.

#zettelkasten #meta

## Related

- [[20200509222116]] Notes guidelines principles

## Contents or Section Title

Fill in with your content!

## References


My biggest fear is that this would become too dispersed and not useful at all.

What do you think?
Do you have any advice?

Post edited by luca on

• @luca, I use my zettelkasten for developing and nurturing both humanities and science knowledge. I'd downvote the folder metaphor as it is too limiting and once you have more than a couple of hundred or a couple thousand notes it becomes a burdensome cognitive weight to think about each time about which is the one folder an idea should be optimally placed in.

If you use this method strictly and only for a narrowly defined scientific field then maybe just maybe folders might work if you could limited yourself to maybe a dozen folders.

And why? With modern full-text search and the ability most apps have of saving searches, it would seem to be a better option to have one idea per note and each note per file and all files in a single folder. I'd much rather search one big bin of notes than have to search several. Also having everything in one bin allows for serendipity, the unexpected collision of ideas which separation into folders limits.

I find tags most useful when I have a project or for labeling a note type (literature, structure, hub) and use them in my saved searches.

I fear that the mix of different type of notes (class notes, my ideas, paper I read will create confusion in the long run)

This is where saved searches and tags really shine. If class notes for MSE393 are all tagged with #MSE393 then it is easy peasy to eliminate the distribution of the other notes with a search on that tag. Don't forget that a Structure Note covering the whole class would also want to be created and maybe some quick access key for during the class duration.

I only occasionally see my whole list of notes in the Notes List. Most of the time I see a filtered list associated in some way with the note I'm working on and am not confused or distracted.

My biggest fear is that this would become too dispersed and not useful at all.

I can report that as my Zettelkasen has become more dispersed (diversified) it has grown in value and become more useful not less.

Will Simpson
I'm a futzing, second-guessing, backtracking, compulsive oversharing, ZK-maniac, in other words, your typical zettelnant.
Research areas: Attention Horizon, Productive Procrastination, Dzogchen, Non-fiction Creative Writing, Cognitive Workload, Python, Data Science
kestrelcreek.com

• The plan looks nice, and I wish you godspeed on your learning journey!

I second @Will's sentiment about folders. If you absolute, 100% certainly want to separate stuff somehow, consider using tags for this first -- depending on the tool you use and how convenient tags are to navigate and to group notes, you can come up with a naming scheme like ##fleeting (double hash) or even #/fleeting (resembling a folder path) or #_fleeting or whatever. Tags can be thought of as "virtual" folders since they delineate subsets of your whole Zettelkasten. The benefit of 1 folder and N tags is that you can more easily connect stuff on the fly without marking it as non-fleeting, moving it to another directory, and/or adding more metadata first. You can just work with a fleeting note as you would with any other, and if it grows, let it grow; and if it withers, so be it!

Author at Zettelkasten.de • https://christiantietze.de/

• Hi @Will @ctietze, thank you very much for your feedback!

I got your point: tags are always better than folders.
I think I agree with you, I will proceed in this way!

Thanks again for sharing your experience!