Effective Tag Management in Information Tools: A Guide for Obsidian and Zotero

This time, I'd like to explore how to use the "tag" feature, which is essential in information management tools.

Classifying saved literature and information using tags is crucial when searching for them later.

However, in practice, many problems arise. People who manage information using tags have likely experienced the following:

• Not knowing how to differentiate between folder and tag functions
• Not knowing what tags to apply in the end
• Having too many tags, making it impossible to find what's where
• Tags overlapping too much, making it unclear where things are located
• Previously saved literature without tags getting buried and becoming difficult to find

I've had plenty of such experiences and failures when using Evernote, Notion, and other tools in the past.

I believe these problems can be solved by reconsidering tags not merely as "search tools" but as "output tools." Today, I'll explore effective tag management methods centered on this concept, with a focus on usage in Obsidian and Zotero.

What Are Tags? The Problem of Distinguishing from Folders

Tags are systems that are typically included in literature and information management software like Obsidian and Zotero. They allow you to attach any number of specific words to a single item, making them convenient for searches and other purposes.

In most software, they differ from folders in that they basically don't take a hierarchical structure. Therefore, they're often used when searching for literature across different categories.

Here, the first problem is distinguishing between folders and tags. At first glance, it might seem that tags are unnecessary if you classify all information into folders, but doing so becomes the first step toward folder hell. In fact, with the literature management software Mendeley that I used 4-5 years ago, folder hell was created, and the situation became impossible to navigate.

Early literature was mostly related to clinical medicine, so I initially thought that classifying by disease would avoid duplication. However, problems frequently arose: Where should reviews dealing with multiple diseases go? How should papers on symptoms and diseases be classified?

The problems with folder structures that become apparent here are the inability to "allow duplication" and the fact that "folders won't be clicked unless their contents can be determined." Much literature and information cannot be classified into a single category and often spans multiple fields, making it impossible to handle such cases.

Given these characteristics, I believe folder structures that don't allow duplication are fundamentally unsuitable for classifications with multiple perspectives (though Zotero does allow some duplication). Also, unless the folder name allows immediate determination of its contents, it's troublesome to click and open each one individually. When searching for files from folders, you end up clicking folders, checking all file names, opening the next one, and so on, so subdivision becomes hellish. The situation becomes even more tragic when hierarchical structures are involved.

For these reasons, I recommend keeping folders to a minimum. Conversely, folder structures are recommended for things that won't change classification and for things where you need to click and immediately see the whole.

Project-based management is often suggested. For example, literature cited in a particular paper, papers used for a specific presentation, and so forth. Opening a folder when working on that project allows you to immediately see the whole, and by looking only at the information there, you can concentrate on your work. Also, once a project is finished, you can keep materials stored in that folder, so when you think "What were those materials I used back then?" later, you can immediately check the contents.

Using tags for output-purpose-based classification and folders for projects would be the most user-friendly approach.

The Problem of What Tags to Apply After All

Now that we understand how to distinguish between tags and folders, the most important issue is "what tags should we apply after all?" Based on guides and past experience, I personally recommend two methods: "envisioning scenarios where tags will be used" and "thinking about tag types when applying them."

Envisioning Scenarios Where Tags Will Be Used

The ultimate purpose of notes and literature organized with tags should be to output them.

• Citing papers → Output as papers
• Citing books → Output as blog articles
• Noting clear explanations → Output as YouTube videos
• Interesting metaphors or symbolic memes → Output as material for presentations

While uses vary by person, choosing what tags to use after considering such specific output purposes should reduce meaningless tagging.

For example, when writing a paper themed "Alzheimer's dementia," classifying by research into categories like "clinical research," "pathology," "biomarkers" would be clear for later reference. Conversely, if this level of classification is already apparent from titles or can be remembered, there's no need to apply tags. How much can be managed by memory alone depends on the number of notes and literature used in the project. I think it's optimal to consider necessary tags according to scale while occasionally organizing and reviewing as mentioned above.

When You Can't Envision Tag Usage Scenarios

There are often cases where you've created notes about literature or information of personal interest but have no immediate plans to output them. In such cases, a method called Feynman's 12 problems might be useful. This is a method described in "Building a Second Brain: A Proven Method to Organize Your Digital Life and Unlock Your Creative Potential" by Tiago Forte. As an aside, almost every book on intellectual production mentions Feynman - everyone really seems to love him.

This method involves keeping 12 problems of personal interest in mind and, when making new discoveries, thinking about which problem's solution they might help with.

It doesn't necessarily have to be 12, but there should be certain trends in topics and problems that interest you. For example, in my case, themes include "medicine," "statistics," "coding," "language learning," "LLM," "philosophy of science," and literature and lectures that interest me generally relate to such themes. While these classifications are quite broad, first classifying into such personal themes and then subdividing if they expand later should prevent literature and notes from being abandoned and reduce forgetting due to untagged items.

Thinking About Tag Types When Applying

Another thing to consider when tagging is thinking about tag types. The following classifications can be considered for tags to apply:

• Classification tags by meaning
• Status management tags
• Tags indicating how to use

First, classification by meaning is probably the most common, involving tags applied based on the content of literature or notes. A statistics paper gets "statistics," a clinical research paper gets "clinical research," and so on - applying tags that match the content. However, these meaning-based tags are particularly prone to wasteful application. You can apply numerous meanings if you want to, and there's a sense of accomplishment from somewhat classifying things. But as mentioned above, tags should be limited to those that need to be used. It's best to think carefully about what meaning you want to filter by when using them.

Next commonly used are status management tags. These indicate the state of items or notes, such as "unread," "reading," "completed" in Zotero. Three stages are simplest, but you might want to add more stages like "skimmed" or "thoroughly read." When creating notes in Obsidian, indicating creation status like "memo," "draft," "completed" might also work. In my case, when creating notes in Obsidian and writing blog articles, I usually finish them immediately, so I don't use multiple status management stages, but in Zotero, I often can't read literature immediately after saving, so I do apply status management like "unread."

The final type of usage tag indicates how to use something. For example, "cite in paper," "reference only, no citation," etc., noting what purpose something will be used for. When writing papers, there are things you directly cite, but also papers you reference for writing Methods or Results. Separating by such purposes makes it clear when you want to look at them again later.

So, summarizing the tagging approach up to this point:

1. Separate things that can be managed by project
2. If subdivision is needed within projects, apply tags envisioning when they'll be used
3. For things not immediately needed, apply tags matching your themes of interest
4. Apply three types: meaning-based tags, status management tags, and usage-indicating tags

This approach should generally keep things to about 3-5 tags and reduce confusion about what to apply.

I think this part is essential and most important, but I'll continue with solutions that can address technical problems.

Too Many Tags, Too Much Tag Overlap Problem

When applying meaning classification tags intuitively without thinking, the problem arises of having too many tags and too much overlap, making it unclear which tag contains what. There are several solutions to this as well.

Aligning Naming Conventions and Definitions

First is aligning rules when naming tags.

For example:

• English or native language?
• For English, singular or plural form? Upper or lowercase?
• Abbreviation or full name?
• Are status-indicating tags clearly defined?

When dealing with literature and information containing many English technical terms, English tag management is often more convenient. Especially for terms without native language translations, translating them yourself can lead to notation variations. Unifying everything in English might be clearer.

Also, while status management tags are recommended, they risk duplication without proper definitions. For example, when classifying by stages like "unread," "skimmed," "thoroughly read," if "thoroughly read," "checked," "completed" etc. coexist, distinctions become unclear. Status tags need to be used with clearly defined personal standards.

Limiting Tag Numbers

While adding as you think of them and organizing later is one approach, I think it's better to keep things minimal while practical necessity remains unclear. Starting with a fixed number like one or two meaning classification tags plus one status tag should prevent tags from becoming enormous.

Zotero actually has an automatic tagging feature, but using such features causes massive tag increases. Keeping things minimal while unclear how much is needed is recommended.

Regularly Reviewing Tags

Like room organization (which I also struggle with), regularly reviewing tags after using them for a while to delete or merge them is probably essential. As initially stated, tags are ultimately tools for outputting something. If what you're outputting is clearly established, tags shouldn't be confusing from start to finish, but whether papers or blogs, what you want to write gradually changes as you write. This is because what you envisioned in your head often feels wrong when written, and discovering different facts while researching is routine. Tags need to change dynamically in response to these changes.

It's good to create a system that ensures regular organization, such as reviewing according to project progress or monthly reviews.

Tools and methods helpful for reviews are discussed later.

Tag Forgetting Problem

Next is the problem of "forgetting to apply tags and not knowing where literature went." For tags, I think the best approach is to apply them when registering items or notes, or at least put them in folders. Without deciding when to apply them, you'll forget.

Obsidian makes tagging easy by just adding #, but Zotero requires clicking the tag button and manual input, making it somewhat troublesome. This is where the numeric key shortcut function becomes useful. In Zotero, right-clicking on created tags reveals a "Color Assignment" command. Here you can set colors to apply to items and numeric key shortcuts (e.g., pressing 1 while an item is selected applies the "unread" tag).

This is very convenient and prevents tag forgetting, so it's recommended. For things you want to read later but are just saving now, applying such "unread" tags is good.

Practical Section: Specific Usage by Tool

Tag Organization in Obsidian

The Tag Wrangler plugin is useful in Obsidian.

This plugin allows:

• Tag merging
• Tag renaming
• Tag deletion (somewhat technical)
• Tag page creation

The first three essentially involve renaming.

For example, right-clicking opens the tag editing window, so you can Rename for easier use, or drag & drop tags to rename them to other tags for merging. Tag deletion is somewhat technical - you press Rename and change to numbers like #1. Since Obsidian doesn't recognize numbers as tags, this effectively serves as tag deletion. However, note that mysterious #1 notations remain in notes.

Also, "Create tag page" or Ctrl+click can create tag description pages. After creation, Ctrl+clicking similarly accesses tag description pages. Creating overviews of articles with tags might also be useful.

When the overall picture is hard to grasp during tag organization, asking Claude or Gemini, which I've introduced before, might be good. They'll detect and overview what content notes exist in what quantities.

However, be careful with these LLMs as they might use data for learning depending on usage, so be cautious with sensitive data.

Tag Organization in Zotero

Zotero's standard tag features make deletion and merging simple. In the library screen's lower left where tags are listed, right-clicking allows deletion or renaming. While there's no merge function, renaming to the same name as other tags automatically merges them.

Also, when wanting to apply tags to multiple items at once, select multiple items with Ctrl or Shift+click and drag & drop to the desired tag in the lower left. This enables batch tagging of items.

I've also summarized similar content about Zotero tags with images in a paid article.

tosuke.gumroad.com/l/zotero_guide

Like Obsidian, LLM usage is also covered in Note articles and previous articles below.

When you can't organize everything yourself, borrowing help from such tools is recommended.

Summary

Today I've explored how tags should be used in Obsidian and Zotero. Tags shouldn't be things you apply and admire, but things that help when you use them. Since usage scenarios are also fluid, it's necessary, though troublesome, to periodically change application methods and organization.

Don't aim for perfection initially - start with simple rules, use them practically, and adjust to match your workflow for success. Let's achieve more efficient information management through output-conscious tagging.