Triples Explained: The Atom of Machine Knowledge
Strip any fact down to its simplest form and you get three parts: something, a relationship, and something else. The pool opens at seven. The hotel is in Pattaya. The restaurant serves Thai food. Each one is a subject, a predicate and an object. In data this three part unit is called a triple, and it is the atom from which knowledge graphs, the Semantic Web, and most structured descriptions of the world are built.
The three parts
The subject is the thing the fact is about. Hotel Azure.
The predicate is the relationship or property. Located in. Has amenity. Opens at.
The object is what the subject is connected to, either another thing or a plain value. Pattaya. Rooftop Pool. 07:00.
Hotel Azure, located in, Pattaya. That is a triple. Hotel Azure, has amenity, Rooftop Pool. Another. Rooftop Pool, opens at, 07:00. A third. Each is small, unambiguous, and complete on its own.
Why three, and not a sentence
A sentence can carry several facts at once and often leaves some implied. "The hotel's rooftop pool opens at seven with great views" contains at least three facts and one opinion, and a machine has to work out where each begins and ends. A triple carries exactly one fact and leaves nothing implied. There is no way to half state it.
That strictness is what makes triples useful. A machine can store them, count them, compare them and check them without ever having to understand a paragraph.
How triples become a graph
The object of one triple can be the subject of another. Hotel Azure, has amenity, Rooftop Pool. Rooftop Pool, opens at, 07:00. Rooftop Pool, has view of, Old Town. The pool appears as an object in the first and as a subject in the next two. Follow those links and you are walking a graph. Enough triples about enough things and you have a knowledge graph, which is nothing more than a large set of triples that share subjects and objects.
This is also why the same fact can be reached from different directions. Starting from the hotel, you find the pool, then its hours. Starting from "things that open at 07:00", you find the pool, then the hotel it belongs to. Nothing has to be duplicated.
Identifiers, not just names
In a serious graph each subject and most objects carry an identifier rather than just a name, so that Pattaya the city and Pattaya the hotel name are never confused, and so that every system using the identifier agrees on what it points to. The name becomes just another triple: this identifier, has name, "Pattaya". That small move is what lets graphs from different sources be combined without arguments about what a word means.
Where you meet triples without noticing
Every structured data block on a website is a bundle of triples: this page's business, has telephone, this number. Every row in a database table is a set of triples sharing one subject. Every knowledge panel in a search result is a display of triples pulled from a graph. And when an AI assistant states a fact about a business, the reliable version of that fact almost always traces back to a triple somewhere.
Why it matters for being understood
If you want a machine to know something about your business, the surest route is to state it as a triple: this thing, this property, this value. Prose can carry the same fact, but it asks the machine to extract it, and extraction is where errors enter. A triple asks nothing. It simply is the fact.
The takeaway
A triple is one fact in its smallest honest form: subject, predicate, object. Chain them and you have a graph. Give them identifiers and they combine across the world. Everything larger in machine knowledge is built from this one small unit.