A page may carry extracted text, a media reference to the rendered page, or both —
the both case is what lets a page be embedded natively by a multimodal model (preserving
tables and charts that text extraction flattens) while its text remains available for
lexical search.
One page of a paginated source (PDF, slide deck).
A page may carry extracted text, a media reference to the rendered page, or both — the both case is what lets a page be embedded natively by a multimodal model (preserving tables and charts that text extraction flattens) while its text remains available for lexical search.