Part 1 · The idea — 04 Two words

What an ontology and a knowledge graph actually are

Two words appear on nearly every page of this site. This page explains them once, in plain language, and adds no further jargon.

01

An ontology is the pattern — it holds no actual things

An ontology is an agreement made in advance: what kinds of things this organisation keeps, what each kind is called, and how the kinds connect to one another.

Say the pattern has objects, and it has makers. Every object goes by a name. Every maker works out of a place. And an object joins a maker through one phrase — made by. Nothing real has been filed against any of it yet.

Written once. Used across the whole organisation.

02

A knowledge graph is that same pattern with the real holdings filled in

Put what you actually hold into the pattern and you have a knowledge graph. One hammered alms bowl, made by the smiths of Ban Bu. The bowl carries its own name, the smiths carry their own place, and every line drawn in the pattern now has something real hanging off it.

Here is the difference that earns its keep. In ordinary systems the link between the bowl and the smiths gets rebuilt from scratch every time somebody asks for it. In a graph that link is already there — it has a name, it has a direction, and anything can walk along it.

THE PATTERN, AND THE SAME PATTERN FILLED IN
THE PATTERN Object Maker made by goes by a name works from a place WHAT YOU ACTUALLY HOLD The alms bowl The Ban Bu smiths made by goes by “Hammered alms bowl” works from Ban Bu, Bangkok
Illustrative. Same shape on both sides. The left one is agreed on before anything is filed; the right one is what the organisation actually holds, sitting in it.
03

Mission-driven — two organisations in the same field should not share one ontology

Two museums can hold the same kinds of objects. But if one exists to teach and the other exists to research, the lines worth drawing are different.

An off-the-shelf ontology therefore serves neither. We design from the organisation's mission, not from an industry template.

04

AI-friendly — it tells the AI what your organisation is for

A general AI guesses from the shape of sentences. An AI working on your knowledge graph follows the lines your own people drew.

So answers trace back to evidence, the same question returns the same answer, and nothing is claimed beyond what the organisation stands behind.

An ontology that matches the mission is what lets AI work for the organisation, rather than in place of it.

Which makes the hard part of this work not technical. It is getting departments to agree on what things are — and only your own people can do that.

08 — Start the conversation

If this is close to what you have been thinking, let's find a time.

90 minutes. Bring your curators or librarians, and one question your institution couldn't answer.

Received. We'll reply within two business days to find a time.

Or write directly — hello@neogens.co

Previous
What it is
Next
MKM for Museums & Libraries