Documents
Going beyond the hit list — previewing a document, extracting its text, finding similar ones, and tagging it.
A search returns records. These pages are about what an interface does with one of them once the user picks it.
Preview
The cached, highlightable rendition of a document, with entity and term highlights addressable by category and by position.
Text chunks
Extracting spans of a document's text by offset — the building block for retrieval-augmented generation.
Similar documents
"More like this", from one document rather than from a query.
Sponsored links
Editorially promoted links, returned outside the engine's ranking.
Labels
User-applied tags, public and private, with their own rights model.
One thing to know before you start
A document preview is rendered by the browser, in an <iframe>, from a URL the server hands you. An
<iframe> cannot carry an Authorization header — it authenticates with the session cookie or not at all.
That is why token-authenticated deployments need the cookie too; see Bearer token.