All projects on Base matching “knowledge-graph”

Search functions or classes by meaning and by keyword. ``mode=bm25`` ranks by lexical match, ``mode=vector`` by embedding similarity to the query text, and ``mode=hybrid`` (default) fuses both rankings with reciprocal rank fusion. Returns up to ``top_k`` matches with their repository, file path and score; the ``domain``, ``ontology_concept_uri``, ``confidence_min`` and ``has_tests`` filters apply

Searches repository names and properties. Returns a paginated envelope with items, total, page, page_size, and total_pages, sorted by composite score.

What should your agent read next? Send text from your corpus; get the concepts most strongly linked to it that it does not yet cover, each with the linking terms and measured link strength. Text is not stored. Text with fewer than 3 known terms is not charged.

Wikidata entity search — find the Wikidata item for a person, organisation, place, work or concept by name or alias, in any language, with id, label, description and which term matched; or find a property by name. Use to get the Q-id before asking for an entity's facts, identifiers or links.