Semantic Web
The Semantic Web is a vision, and a set of W3C standards, for extending the World Wide Web so that information carries machine-interpretable meaning, allowing software to find, combine, and reason over data published by different sources rather than only displaying pages to humans. The ordinary web links documents; the Semantic Web links statements about things. A page saying that a person was born in a given city becomes a structured assertion that any program can consume, merge with other assertions, and query, without scraping prose.
The vision was popularized by Tim Berners-Lee, James Hendler, and Ora Lassila in the May 2001 Scientific American article "The Semantic Web," which pictured software agents arranging appointments and services by exchanging structured data with one another. The underlying idea was older: Berners-Lee had described a web of data and meaning in his earlier architectural notes, and the W3C had already published the first RDF specification in 1999. The 2001 article gave the effort its public narrative, and with it a long period of both investment and skepticism about whether publishers would ever annotate their content with the required care.
The technical stack rests on a deliberately simple data model, RDF, in which every fact is a triple of subject, predicate, and object, with things and relations named by URIs. The statement that one resource was written by another is one triple, and a set of triples forms a graph that can be merged with any other graph simply by taking the union. RDF Schema adds basic classes and subclass relations, and OWL, a W3C recommendation since 2004 and revised as OWL 2 in 2009, adds description-logic axioms for richer inference. SPARQL, standardized in 2008 and extended as SPARQL 1.1 in 2013, queries these graphs by matching triple patterns, in the way SQL queries tables. Serializations such as Turtle and JSON-LD, the latter a W3C standard from 2014, make the data practical to write and embed in web pages.
Linked Data was the pragmatic turn that gave the project its real-world traction. In 2006 Berners-Lee set out four principles: use URIs as names for things, make them dereferenceable over HTTP, return useful RDF when they are looked up, and include links to other URIs so more can be discovered. The Linking Open Data effort that began in 2007 connected datasets such as DBpedia, which extracts structured data from Wikipedia, and grew into a large cloud of interlinked public datasets. Wikidata, launched in 2012 by Wikimedia, is a collaboratively edited knowledge base exposed through RDF and SPARQL. In 2011 Google, Microsoft, Yahoo, and Yandex launched schema.org, a shared vocabulary for marking up pages, and Google's May 2012 announcement of its Knowledge Graph, with the slogan "things, not strings," brought the idea of entity-centered graphs to mainstream search.
The grand vision of autonomous agents reasoning across the open web mostly did not arrive, and the reasons are instructive. Formal inference proved fragile on messy, inconsistent, and sometimes adversarial data; authors had little incentive to publish rich markup; and agreeing on shared vocabularies is a social problem more than a technical one. A common misconception is therefore that the Semantic Web either triumphed completely or failed completely, when the more accurate picture is partial adoption: structured markup and knowledge graphs are everywhere behind search and enterprise data integration, while the open, reasoning-driven web of agents remains largely aspirational. Another confusion is equating the Semantic Web with knowledge graphs in general, since many industrial graphs use property-graph databases with no RDF at all.
Modern language models revive the questions the project posed. Models extract entities and relations from prose at scale, which attacks the annotation bottleneck, while graphs supply verifiable facts that retrieval-augmented systems can use to ground answers. Standards for provenance, identifiers, and shared schemas remain the hard part, because a model that guesses a relation is not publishing a trustworthy statement.
Read the Semantic Web as the web's address-and-link idea applied to facts rather than pages: triples with shared names let independent datasets merge automatically, and its lasting legacy is the knowledge graph and structured markup that quietly underlie search, rather than the fully autonomous reasoning agents promised in 2001.