RDF, OWL and SPARQL: The Semantic Web in Ten Minutes
Long before large language models, a group of researchers led by the inventor of the web set out to make web content readable by machines, not just by people. They called it the Semantic Web, and they built three standards that are still the backbone of serious knowledge graphs today. RDF is how facts are written. OWL is how the rules about those facts are defined. SPARQL is how you ask questions. Here is what each one does, without the jargon.
RDF: facts as three part sentences
RDF stands for Resource Description Framework. The idea is simple. Every fact is a sentence with three parts: a subject, a predicate and an object. Hotel Azure, located in, Pattaya. Hotel Azure, has amenity, Rooftop Pool. Rooftop Pool, opens at, 07:00. These three part facts are called triples.
The clever part is that every piece of a triple can be a web address. Instead of the word "Pattaya", which could mean the city or a hotel named after it, RDF uses a unique identifier, something like a URL, that points to exactly one thing. Two systems that both use that identifier are talking about the same Pattaya, without any negotiation.
A knowledge graph in RDF is just a very large collection of triples, where the objects of some triples are the subjects of others. That is what links them into a graph.
OWL: the rules about the facts
OWL is the Web Ontology Language. Where RDF lets you state facts, OWL lets you state what kinds of facts are allowed and what they imply. It is how you write an ontology in a form a machine can enforce.
In OWL you can say: Hotel is a class. Restaurant is a class. "Has outlet" is a relationship whose subject must be a Hotel and whose object must be an Outlet. Restaurant is a kind of Outlet. Every Hotel is located in exactly one City.
Once those rules exist, a machine can reason. If Hotel Azure has outlet Salt, and Salt is a Restaurant, and Restaurant is a kind of Outlet, the machine concludes that Salt is an Outlet without anyone writing that down. If someone tries to record a Hotel located in two cities, the machine can flag it. OWL turns a pile of facts into something that checks itself.
SPARQL: asking the graph a question
SPARQL is the query language for RDF, in the way SQL is the query language for tables. Instead of selecting rows from a table, you describe a pattern of triples and ask the graph for everything that matches.
In plain words a SPARQL query might read: find every restaurant that is an outlet of a hotel located in Pattaya, where the restaurant closes after 22:00 and allows children. The query walks the graph following those relationships and returns the restaurants that fit. Because the graph is made of named relationships, questions that span several hops are natural rather than painful.
Why these matter in the AI era
For a while the Semantic Web looked like an academic project. Then two things happened. Google built its knowledge graph on exactly these ideas. And language models arrived, brilliant at prose and unreliable at facts, needing a source of truth they could not blur.
RDF gives that source of truth a standard shape. OWL keeps it consistent. SPARQL lets a system fetch precisely the facts a question needs. GraphRAG, structured data markup, and most serious enterprise knowledge graphs are built on this stack or on something that maps directly onto it.
The takeaway
RDF writes facts as triples with unambiguous identifiers. OWL defines the types and rules those facts must follow. SPARQL asks questions by matching patterns across the graph. Together they are how the web was meant to become readable by machines, and they are the quiet foundation under most of the systems now deciding what an AI knows about you.