Enum: QuoteRoleEnum
Where the quoted sentence sits in the cited publication's own argument: a finding that publication produced, something it restates from elsewhere, or its synthesis of a literature it did not generate.
URI: dismech:enum/QuoteRoleEnum
Permissible Values
| Value | Meaning | Description | Additional Info |
|---|---|---|---|
| PRIMARY_RESULT | None | The quoted text reports an observation, measurement, analysis, or conclusion ... | Title: Primary result |
| BACKGROUND | None | The quoted text restates something established elsewhere -- an introduction, ... | Title: Background |
| REVIEW_SYNTHESIS | None | The quoted text is the cited publication's synthesis of work it did not itsel... | Title: Review synthesis |
Slots
| Name | Description |
|---|---|
| quote_role | Where the quoted sentence sits in the cited publication's own argument -- a r... |
Comments
- This is a fact about the cited document, not a judgement about the claim, and not a strength grade. It is orthogonal to all three existing evidence axes and composes with each of them:
supportsis direction,directnessis inferential distance,evidence_sourceis study type. A sentence can be BACKGROUND and still be DIRECT, SUPPORT and HUMAN_CLINICAL -- that combination is precisely the case issue #10262 was filed about. - The motivating case (#10262): a chick-embryo study whose introduction states the human clinical picture, quoted for that human fact. Grading it MODEL_ORGANISM asserts a chick measured human perinatal mortality; grading it HUMAN_CLINICAL asserts a study type the paper never ran; the OTHER that review settled on says nothing at all. With
quote_rolethe item is HUMAN_CLINICAL + BACKGROUND, which is both true and queryable --evidence_sourcegrades the evidence the quoted text describes, andquote_rolerecords that the citing paper is not where it came from. - Every value takes a
close_mappingsto CiTO rather than anexact_mappings, because the two vocabularies describe different ends of the same citation. A CiTO property types the citing entity's use of a reference;quote_rolerecords where the sentence sits inside the cited document. They correlate and they come apart: a dismech item quoting an introduction sentence as support for a KB claim iscitesAsEvidencefrom the citing side while itsquote_roleis BACKGROUND. Assertingexact_mappingswould claim an equivalence that does not hold in that case, so all three are close matches and the mapping stays usable for an exporter without overstating it. - Deliberately has no UNKNOWN value, unlike DirectnessEnum. Absent already means "nobody has assessed this", which is the honest state of every evidence item predating the slot; a second spelling of the same thing would only invite curators to fill one in. CLAUDE.md gives the same instruction for
directness, where UNKNOWN is a legacy value. - Partly derivable, which is the test docs/explanation/evidence-model.md sets for an authored slot ("strength stays derived, not authored"): where a reference's cached body carries NLM structured-abstract section labels, locating the snippet between two of them decides the zone without inference.
just list-background-citationsreports that derivation as a worklist. It is a worklist and not an autofill: it places roughly a quarter of PMID-cited snippets (the rest are quoted from full-text prose that carries no labels), a structured BACKGROUND: paragraph routinely closes with the authors' own framing, and the same line dismech-terms draws for ontology-term suggestions applies here.
Identifier and Mapping Information
Schema Source
- from schema: https://w3id.org/monarch-initiative/dismech
LinkML Source
name: QuoteRoleEnum
description: 'Where the quoted sentence sits in the cited publication''s own argument:
a finding that publication produced, something it restates from elsewhere, or its
synthesis of a literature it did not generate.'
comments:
- 'This is a fact about the cited document, not a judgement about the claim, and not
a strength grade. It is orthogonal to all three existing evidence axes and composes
with each of them: `supports` is direction, `directness` is inferential distance,
`evidence_source` is study type. A sentence can be BACKGROUND and still be DIRECT,
SUPPORT and HUMAN_CLINICAL -- that combination is precisely the case issue #10262
was filed about.'
- 'The motivating case (#10262): a chick-embryo study whose introduction states the
human clinical picture, quoted for that human fact. Grading it MODEL_ORGANISM asserts
a chick measured human perinatal mortality; grading it HUMAN_CLINICAL asserts a
study type the paper never ran; the OTHER that review settled on says nothing at
all. With `quote_role` the item is HUMAN_CLINICAL + BACKGROUND, which is both true
and queryable -- `evidence_source` grades the evidence the quoted text describes,
and `quote_role` records that the citing paper is not where it came from.'
- 'Every value takes a `close_mappings` to CiTO rather than an `exact_mappings`, because
the two vocabularies describe different ends of the same citation. A CiTO property
types the *citing* entity''s use of a reference; `quote_role` records where the
sentence sits inside the *cited* document. They correlate and they come apart: a
dismech item quoting an introduction sentence as support for a KB claim is `citesAsEvidence`
from the citing side while its `quote_role` is BACKGROUND. Asserting `exact_mappings`
would claim an equivalence that does not hold in that case, so all three are close
matches and the mapping stays usable for an exporter without overstating it.'
- Deliberately has no UNKNOWN value, unlike DirectnessEnum. Absent already means "nobody
has assessed this", which is the honest state of every evidence item predating the
slot; a second spelling of the same thing would only invite curators to fill one
in. CLAUDE.md gives the same instruction for `directness`, where UNKNOWN is a legacy
value.
- 'Partly derivable, which is the test docs/explanation/evidence-model.md sets for
an authored slot ("strength stays derived, not authored"): where a reference''s
cached body carries NLM structured-abstract section labels, locating the snippet
between two of them decides the zone without inference. `just list-background-citations`
reports that derivation as a worklist. It is a worklist and not an autofill: it
places roughly a quarter of PMID-cited snippets (the rest are quoted from full-text
prose that carries no labels), a structured BACKGROUND: paragraph routinely closes
with the authors'' own framing, and the same line dismech-terms draws for ontology-term
suggestions applies here.'
from_schema: https://w3id.org/monarch-initiative/dismech
rank: 1000
permissible_values:
PRIMARY_RESULT:
text: PRIMARY_RESULT
description: The quoted text reports an observation, measurement, analysis, or
conclusion the cited publication itself produced.
title: Primary result
close_mappings:
- cito:citesAsEvidence
BACKGROUND:
text: BACKGROUND
description: The quoted text restates something established elsewhere -- an introduction,
a background or framing sentence, a motivation for the work -- rather than a
finding of the cited publication.
title: Background
close_mappings:
- cito:obtainsBackgroundFrom
REVIEW_SYNTHESIS:
text: REVIEW_SYNTHESIS
description: 'The quoted text is the cited publication''s synthesis of work it
did not itself perform: a review, commentary, editorial, or consensus or guideline
statement summarizing a literature.'
title: Review synthesis
comments:
- The furthest of the three from its CiTO neighbour. `citesAsAuthority` is the
nearest property but carries an "uncontested authority" reading that REVIEW_SYNTHESIS
does not assert -- the value records what kind of document the sentence sits
in, not how much weight to give it.
close_mappings:
- cito:citesAsAuthority