Taxonomies

A hierarchical classification system that organises things into categories and subcategories --- from the most general at the top to the most specific at the bottom.

---

What is it?

A taxonomy is a system for classifying things into a hierarchy. At the top sit broad categories. Each level down gets more specific, until you reach individual items at the bottom. The word comes from the Greek taxis (arrangement) and nomos (law) --- literally, the laws of arrangement.¹

You already use taxonomies constantly, even if you have never used the word. Every folder structure on your computer is a taxonomy. Every menu in a restaurant (Starters > Soups > French Onion) is a taxonomy. Every library classification system is a taxonomy. The pattern is always the same: general to specific, with each item belonging to exactly one parent category.²

In the context of knowledge-graphs, a taxonomy is a specific type of graph structure --- one where every edge represents “is a kind of” or “belongs to” and every node has at most one parent. This makes it a tree (or more precisely, a directed acyclic graph where each node has a single parent). A knowledge graph can contain many types of relationships; a taxonomy restricts itself to hierarchical containment.³

This restriction is a feature, not a limitation. By constraining the structure to a strict hierarchy, a taxonomy makes classification unambiguous: every item has exactly one place. You can always answer “where does this belong?” by walking up the tree to its parent, grandparent, and so on until you reach the root.

In plain terms

A taxonomy is like a set of Russian nesting dolls (matryoshka). The biggest doll contains a smaller one, which contains a smaller one still. Each doll fits inside exactly one larger doll, and the arrangement tells you how specific something is --- the smallest doll is the most precise category.

---

At a glance

A taxonomy as a tree structure (click to expand)

graph TD
    A[Living Things] --> B[Animals]
    A --> C[Plants]
    B --> D[Mammals]
    B --> E[Birds]
    D --> F[Dogs]
    D --> G[Cats]
    E --> H[Eagles]
    E --> I[Penguins]

Key: Each node is a category. Each edge means “contains” or “is a parent of.” Every node has exactly one parent (except the root). The further down you go, the more specific the classification becomes. This is the defining structure of a taxonomy: a tree where specificity increases with depth.

---

How does it work?

1. Levels of specificity

A taxonomy is organised into levels, where each level represents a degree of specificity. The topmost level is the broadest category; the bottommost is the most precise.¹

For example, the Linnaean biological taxonomy uses eight major ranks:¹

Level	Example (red fox)
Domain	Eukarya
Kingdom	Animalia
Phylum	Chordata
Class	Mammalia
Order	Carnivora
Family	Canidae
Genus	Vulpes
Species	Vulpes vulpes

The same principle applies outside biology. A knowledge system might use levels like: domain > discipline > topic > concept > atomic. A library uses: broad subject > narrow subject > author > title. The number and names of levels vary, but the pattern is always the same: each level is a refinement of the one above.

Think of it like...

Levels in a taxonomy are like zoom levels on a map. At the highest zoom-out, you see continents. Zoom in and you see countries. Zoom further and you see cities, then neighbourhoods, then individual buildings. Each zoom level reveals more detail while still fitting inside the boundaries of the level above.

2. Parent-child relationships

The fundamental relationship in a taxonomy is parent-child: a parent category contains one or more child categories. Each child has exactly one parent (in a strict taxonomy). This creates a tree structure, which is a special case of the directed graphs covered in nodes-and-edges.²

The parent-child constraint means there is always exactly one path from any node to the root. This is what makes classification unambiguous --- you never have to decide “which of my two parents do I belong to?”

Think of it like...

A family tree where every child has exactly one parent (simplified, of course). You can always trace a single line upward from any person to the common ancestor at the top. There are no forks going upward --- just one clear lineage.

Key distinction

A taxonomy has single-parent hierarchy (a tree). A knowledge graph allows multiple parents and cross-cutting relationships (a general graph). Taxonomies trade flexibility for clarity: you lose the ability to say “this concept belongs to two categories at once,” but you gain unambiguous classification.

3. Classification rules

For a taxonomy to be useful, it needs consistent classification rules --- the criteria for deciding which category something belongs to. Without clear rules, two people might classify the same item differently, making the taxonomy unreliable.³

Good classification rules are:

• Mutually exclusive --- an item belongs to exactly one category at each level
• Collectively exhaustive --- every item can be classified somewhere
• Consistent --- the same criteria apply at each level

Example: a recipe taxonomy (click to expand)

Consider classifying recipes:

Level	Categories	Classification criterion
Cuisine	Italian, Japanese, Mexican, etc	Country of origin
Course	Starter, Main, Dessert	When it is served
Method	Baked, Fried, Raw	How it is prepared
Dish	Margherita pizza, Tempura, Ceviche	Specific recipe

The rule at each level is clear: cuisine is determined by origin, course by timing, method by preparation technique. A Margherita pizza is Italian > Main > Baked > Margherita pizza. The path is unambiguous.

But notice the tension: what if a dish is both a starter and a main? Strict taxonomies force a choice. This is the fundamental trade-off of hierarchical classification.

4. Flat tags vs hierarchical taxonomies

A common alternative to a taxonomy is flat tagging --- applying labels to items without any hierarchy. For example, tagging a blog post with “python”, “tutorial”, “beginner” gives you three independent labels with no structure between them.⁴

Tags are flexible but unstructured. They cannot express “python is a programming language” or “a tutorial is a type of content.” A taxonomy can, because the hierarchy encodes those relationships. The trade-off is that tags are easier to create (just type a word), while taxonomies require upfront design.⁴

In practice, many systems use both: a taxonomy for the primary classification (where does this belong?) and tags for secondary attributes (what else is it about?).

Think of it like...

Tags are like sticky notes on a book --- you can put as many as you want, in any combination, but they do not tell you where the book goes on the shelf. A taxonomy is the shelf system itself: every book has one specific place, determined by its classification.

---

Why do we use it?

Key reasons

1. Enabling retrieval. A well-designed taxonomy lets you find things by navigating from general to specific. If you know you want a “mammal” but not which one, you can browse the Mammals branch and discover options you did not know existed. Flat lists do not support this kind of exploratory navigation.²

2. Resolving prerequisites. In a learning system, a taxonomy encodes what must be understood before tackling something more specific. If “Calculus II” is a child of “Calculus I” in the taxonomy, the prerequisite relationship is built into the structure itself.³

3. Creating shared vocabulary. A taxonomy forces an organisation to agree on what categories exist and what they mean. This eliminates the problem of different teams using different names for the same thing --- a pervasive issue in large organisations.⁴

4. Supporting automation. When a taxonomy is encoded in a machine-readable format, software can classify new items automatically, generate navigation menus, build breadcrumb trails, and recommend related content --- all by traversing the tree structure.

---

When do we use it?

• When you need to organise a growing collection of items (documents, products, concepts) into a navigable structure
• When users need to browse by category rather than search by keyword
• When a system requires prerequisite ordering --- knowing what must come before what
• When multiple people or systems need a shared classification to communicate consistently
• When building navigation, breadcrumbs, or faceted search for a website or application

Rule of thumb

If you find yourself creating the same set of tags repeatedly, or if users cannot find things because there are too many items in a flat list, you need a taxonomy.

---

How can I think about it?

The department store

A department store is a physical taxonomy.

• The store is the root node
• Each floor is a top-level category (Clothing, Electronics, Home)
• Each section on a floor is a subcategory (Clothing > Women > Dresses)
• Each rack is a further refinement (Dresses > Evening > Size 10)
• The garment is the leaf node --- the most specific item

You navigate by narrowing: pick a floor, then a section, then a rack. You never have to search the entire store. The hierarchy does the work of filtering for you. And if you ask “where are the evening dresses?”, there is exactly one answer: Floor 2, Women, Dresses, Evening.

The postal address

A postal address is a taxonomy of location.

• Country is the broadest level (Switzerland)
• Canton/State narrows it (Vaud)
• City narrows further (Lausanne)
• Street is more specific (Rue de Bourg)
• Building number is the leaf (14)

Each level contains the next. Every address traces a single path from country to building. Two people at the same address share the same path; two people in different cities diverge at the city level. The hierarchy makes sorting, routing, and delivery efficient --- a postal worker does not need to know every address in the country, just how to narrow down from the top.

---

Concepts to explore next

Concept	What it covers	Status
topological-sort	Ordering nodes so that dependencies come before dependents --- used to resolve prerequisite chains in taxonomies	stub
structured-data-vs-prose	The trade-off between hierarchical structure and free-form text	stub
databases	How taxonomies and other structures are stored and queried at scale	complete

Some cards don't exist yet

A broken link is a placeholder for future learning, not an error.

---

Check your understanding

Test yourself (click to expand)

1. Explain what a taxonomy is to someone who has never used the term. Use an everyday example.
2. Name three real-world taxonomies and identify the levels of specificity in each.
3. Distinguish between a taxonomy and a set of flat tags. What can a taxonomy express that tags cannot?
4. Interpret this scenario: a knowledge system has a concept classified under two different parent categories. What problem does this create, and how would a strict taxonomy resolve it?
5. Connect taxonomies to knowledge graphs: in what sense is a taxonomy a special case of a knowledge graph, and what does it give up in exchange for that simplicity?

---

Where this concept fits

Position in the knowledge graph

graph TD
    KG[Knowledge Graphs] --> NE[Nodes and Edges]
    KG --> TAX[Taxonomies]
    KG --> TS[Topological Sort]
    TAX -.->|related| TS
    TAX -.->|related| SDP[Structured Data vs Prose]
    style TAX fill:#4a9ede,color:#fff

Related concepts:

• topological-sort --- taxonomies with prerequisites can be topologically sorted to produce a valid learning order
• structured-data-vs-prose --- taxonomies are a form of structured data; the trade-off between structure and flexibility is a recurring design decision
• databases --- taxonomies are often stored in relational databases (as adjacency lists) or graph databases (as parent-child edges)

---

Sources

---

Further reading

Resources

• Taxonomy: the science of classification (Royal Belgian Institute of Natural Sciences) --- Accessible introduction to biological taxonomy with clear explanations of why classification matters
• Taxonomy Hierarchy Levels (The Accidental Taxonomist) --- Practical guide to designing taxonomy levels for information systems, not just biology
• The Taxonomic Classification System (Lumen Learning) --- Beginner-friendly walkthrough of the Linnaean system with diagrams
• What is taxonomy? (Natural History Museum) --- Beautiful overview from one of the world’s leading natural history institutions

Footnotes

1. Wikipedia contributors. (2024). Taxonomic rank. Wikipedia. ↩ ↩² ↩³

2. Lumen Learning. (n.d.). The Taxonomic Classification System. Biology for Majors I. ↩ ↩² ↩³

3. Hedden, H. (2018). Taxonomy Hierarchy Levels. The Accidental Taxonomist. ↩ ↩² ↩³

4. Natural History Museum. (2025). What is taxonomy?. Natural History Museum, London. ↩ ↩² ↩³