authorities

Citation Phrase Matches

Return the subgraph edges whose citing paragraph contains phrase.

Two steps, mirroring the extracts UX's phrase semantics:

  1. The Meili paragraphs index (quoted → true phrase match) yields the (authority_id, paragraph_ref) paragraphs that contain the phrase, scoped to the citing authorities in the node set. The per-chunk search is split adaptively (:func:_phrase_matched_paragraphs) so a common phrase matching more paragraphs than the index cap is never silently truncated.
  2. Those are intersected with each edge's own citing paragraph_ref:
    • a pinpoint edge lights only when its exact paragraph carries the phrase, so a phrase in some other paragraph never lights it;
    • a non-pinpoint edge (paragraph_ref = "0" — the cover/preamble sentinel, whose short preamble is often not indexed as its own chunk) lights when the phrase appears anywhere in the citing case, which is the most we can assert without a real pinpoint.

Known limitation: paragraph_ref is not unique within a case for restart-numbered judgments (ET/FTT renumber from 1 in each JUDGMENT/REASONS section). An edge whose ref collides across sections lights if either same-numbered paragraph carries the phrase — the stored ref alone cannot say which section the citation is in. Fixing this precisely needs a section-qualified ref recorded at ingest, not a query-time change.

Returns {"pairs": [[citing_id, cited_id], ...]} — the same id pairs the client already keys edges by (from→to).

post/authorities/graph/citation-phrase-matches

Request

  • Base URL: https://api.casenode.ai
  • URL: https://api.casenode.ai/authorities/graph/citation-phrase-matches
  • Auth: none declared

Request body

idsstring[]
phrasestring

Example request

{
  "ids": [
    "uk.eat.2022.68",
    "uk.uksc.2012.37",
    "uk.ewca.civ.2007.33"
  ],
  "phrase": "burden of proof"
}

Response

Successful Response

object required

Changes

No recorded changes to this operation across all 1 revision of this API.