Semantic Web and Ontologies

The Semantic Web is an effort to make information on the internet machine-readable not just by labeling data, but by encoding the relationships and meanings that connect it — so that a computer can reason about the fact that a "physician" is a kind of "person" who "treats" "patients," rather than merely matching strings of text. Researchers use formal languages like RDF and OWL to build ontologies, which are structured vocabularies that define concepts and their logical relationships within a domain, and SPARQL to query the resulting networks of linked data across distributed sources. A central challenge is getting independently designed systems to agree on shared meaning — the schema matching problem — since two databases might use different terms for the same idea or the same term for different ideas. Active work focuses on scaling reasoning over massive knowledge graphs, automatically learning ontologies from text, and integrating these structured representations with neural language models that are powerful but largely opaque about what they actually "know."

Works

171,740

Total citations

1,492,321

Keywords

Semantic WebOntologyLinked DataRDFOWLSchema Matching