DisamKB
A large general-domain knowledge base entirely from a large language model.
Browse all Dataset profile SPARQL query
38.4 million triples 1.6 million entities
96% entity precision 100% homonymy 91% synonymy resolved
About
DisamKB is a general-domain knowledge base built from GPT-5 — 38.4 million triples across 1.6 million disambiguated entities. The browser exposes the disambiguation chain (surface form → canonical entity → provenance batch) on every triple.