“Ontology” is a word that tends to clear a room. It sounds like something that needs a committee, a standards body and several years of meetings. It’s the kind of thing large organizations build to make their databases talk to each other, and individuals politely avoid.
Here’s the contrarian claim: the person who most needs a small ontology is you, working alone, in your own notes.
Not a formal, enterprise-grade one. A small, deliberate vocabulary for the kinds of things you think about and the ways they relate. Because the main problem with most personal knowledge systems isn’t that they’re too disorganized. It’s that the same idea gets written down five different ways, and the system has no way of knowing they’re the same idea.
What problem does an ontology solve for one person?
Consider how notes accumulate. In March you write about a “client.” In June you call the same kind of thing a “customer.” In September it’s an “account.” You tag one note #clients, another #customers. A project note mentions “the Henderson engagement” without linking to anything.
Each note made sense when you wrote it. Collectively, they’re a vocabulary drift. Search for “client” and you miss two-thirds of what you know. Try to see everything related to one engagement and you’re reconstructing it from memory.
An ontology solves this by settling, once, three things:
- What kinds of things exist in your work (concepts or classes, like Client, Project, Decision).
- How they relate (relations, like “Project forClient Client”).
- Which specific things are instances of each kind (the Henderson engagement is a Project).
That’s it. A shared vocabulary, shared between you in March and you in September.
Hasn’t this been tried before?
For centuries, in various forms. Three useful precedents:
Locke’s commonplace book. In 1685, John Locke published a method for indexing a commonplace book, the personal notebooks in which readers copied passages worth keeping. His system used a compact index keyed to the first letter and first vowel of a heading. The point wasn’t elegance. It was retrieval: a commonplace book without a consistent heading scheme was a pile.
Vannevar Bush’s memex. In his 1945 essay “As We May Think,” Bush imagined the memex, a desk-sized device for storing a person’s books, records and communications, linked by “associative trails” the user created. He argued that the human mind works by association, and a personal record should too.
Luhmann’s Zettelkasten. The German sociologist Niklas Luhmann kept a slip-box of index cards over decades, each with a fixed identifier and links to related cards. He credited the system with much of his prolific output, describing it almost as a conversation partner.
Each solved a version of the same problem: how to make a personal collection of knowledge retrievable and connectable. An ontology is the modern, explicit version of what each of them did: decide what kinds of things you keep and how they connect.
Why not just use tags?
Tags are useful, and fast, and they drift. A tag is a label with no definition, no relations and no rules. #client and #clients and #customer coexist happily, and nothing stops a fourth.
An ontology adds three things tags lack:
- Definitions. Each concept has a note saying what it is and isn’t. “Client: an organization or person who pays for work. Not a prospect.”
- Relations. Things connect in named ways. A Decision affects a Project. A Project is forClient a Client. Now you can follow the connections.
- Types. Each note is an instance of something. When you open a Project note, you know what fields to expect.
Tags tell you a note is about something. An ontology tells you what the note is and what it’s connected to.
What does a good concept definition look like?
Most of the value of a personal ontology lives in the definitions, and most definitions are too vague to help. “Project: a project” is a placeholder, not a definition. A useful one answers four questions:
- What is it? “A Project is a bounded piece of work with a defined outcome and an end.”
- What isn’t it? “Not an ongoing responsibility. Ongoing work is an Area.”
- How do you recognize one? “It has a deliverable and a date, even an approximate one.”
- What does it connect to? “Every Project is forClient a Client, and may have Decisions that affect it.”
Here’s how that looks for a few common concepts:
| Concept | Is | Isn’t |
|---|---|---|
| Client | Someone who pays for work | A prospect who hasn’t paid yet |
| Source | Something you read or heard and may cite | Your own idea about it |
| Decision | A choice made, with a reason and a date | An open question still being discussed |
The “isn’t” column does the heavy lifting. It’s where drift gets stopped. When you hesitate over whether a new note is a Source or an Idea, the definition decides for you, and it decides the same way every time.
What happens when a definition stops fitting?
Change it, deliberately. Write a short Decision note recording what changed and why, then update the definition. Your ontology isn’t a monument. It’s a working agreement, and working agreements get revised. The Decision note means that six months from now you’ll know why “Client” suddenly includes retainer partners, instead of wondering whether past-you was confused.
Doesn’t this add a lot of overhead?
Less than people expect, because you only define what you actually use. A personal ontology might have ten concepts and a handful of relations. It grows when your work demands it, not before.
And it removes overhead elsewhere. When every Decision note uses the same template, you stop reinventing structure. When every Project links to its Client, you stop hunting. The upfront cost is one afternoon of deciding. The ongoing benefit is never having to decide again.
What does a small personal ontology look like?
The Ontology & Knowledge-System Starter Kit teaches the structure through a fictional Home Library example, deliberately simple:
- Concepts: Book, Author, Shelf, Genre.
- Relations: writtenBy (Book → Author), locatedOn (Book → Shelf), hasGenre (Book → Genre).
- Instances: specific books, authors and shelves, each as its own note.
That’s a complete, working ontology. It’s small enough to grasp in a minute, and it demonstrates every piece: classes, relations and instances. Once you see it, you can sketch your own for clients, projects, sources, ideas or anything else you work with.
How do layers keep it from getting messy?
The kit organizes the vault into three layers:
- Core: the general building blocks, such as what a concept, a relation or a decision is.
- Domain: the concepts specific to your field or work, like Client or Project.
- Applied: the instances and day-to-day notes that use the vocabulary.
Layers keep the vocabulary stable while the notes multiply. You can add hundreds of instance notes in the Applied layer without touching the definitions in Domain, and you can extend Domain without disturbing Core.
How do you start without boiling the ocean?
Six steps, in an afternoon:
- List the nouns you write about most. Clients, projects, sources, meetings, ideas, whatever recurs.
- Pick three to five that cause the most retrieval pain.
- Write a one-paragraph definition for each, including what it isn’t.
- Name two or three relations between them.
- Convert ten existing notes into instances that use the new vocabulary.
- Live with it for a week, then adjust definitions based on what felt awkward.
Resist the urge to model everything. A small ontology that you actually use beats a beautiful one you abandon in week two.
Where does this differ from a content inventory graph?
A content inventory graph models a website’s pages and how they relate, so a business can see gaps and connections in what it publishes. A personal ontology models your thinking: the concepts, decisions and projects in your own work. The underlying ideas are similar, classes, relations and instances, but the purpose is different. One is for an audience. This one is for you.
What’s in the kit?
The Ontology & Knowledge-System Starter Kit is an Obsidian vault organized into Core, Domain and Applied layers, with eight templates: Concept/Class, Relation, Instance, Hub, Decision, Digest, Project Workbench and Daily. The fictional Home Library example is included as linked notes, as a Turtle (.ttl) file for anyone who wants to open it in a formal tool, and as a CSV. A plain-language guide walks through the ideas.
No plugins are required. Protégé is optional if you want to explore the formal side.
Quick answers
Do I need to know RDF or OWL?
No. The vault works as plain Markdown notes. The Turtle file is there if you want to explore further.
Does it work outside Obsidian?
The notes are Markdown, so they’re readable anywhere. The linking experience is best in Obsidian or a similar tool.
How big should my ontology get?
As big as your work requires and no bigger. Most personal ontologies stay small.
A vocabulary you share with yourself
Every note you write is a message to your future self. An ontology makes sure you’re both speaking the same language: the same words for the same things, connected in the same ways.
Get the Ontology & Knowledge-System Starter Kit and spend one afternoon settling the vocabulary of your own work. September you will be grateful to March you.
The starter kit lives on Tools for Obsidian, where we keep the vaults and templates we build for that app.
Leave a Reply