Class: ExtractionMetadata¶

Pipeline provenance stamped onto the extraction record at generation time.

 classDiagram
    class ExtractionMetadata
    click ExtractionMetadata href "../ExtractionMetadata/"
      ExtractionMetadata : extraction_date

      ExtractionMetadata : extractor_model

      ExtractionMetadata : extractor_version

      ExtractionMetadata : paper_sections





        ExtractionMetadata --> "*" PaperSection : paper_sections
        click PaperSection href "../PaperSection/"



      ExtractionMetadata : repaired_by

      ExtractionMetadata : source_text_hash

Slots¶

Name	Cardinality and Range	Description	Inheritance
extractor_model	1 String	Model identifier used for extraction, e	direct
extractor_version	1 String	Version string of the extraction prompt or pipeline	direct
source_text_hash	0..1 String	SHA-256 of the paper text processed, for reproducibility checking	direct
extraction_date	0..1 Date	ISO-8601 date of extraction	direct
repaired_by	0..1 String	Identifier of the repair pass that edited this record after extraction, absen...	direct
paper_sections	* PaperSection	Complete deterministic index of all paper sections and subsections, including...	direct

Usages¶

used by	used in	type	used
Study	extraction_metadata	range	ExtractionMetadata

Identifier and Mapping Information¶

Schema Source¶

from schema: https://neurostuff.org/schemas/neuroimaging-study-extraction

Mappings¶

Mapping Type	Mapped Value
self	neurostudy_ex:ExtractionMetadata
native	neurostudy_ex:ExtractionMetadata

LinkML Source¶

Direct¶

name: ExtractionMetadata
description: Pipeline provenance stamped onto the extraction record at generation
  time.
from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction
rank: 1000
attributes:
  extractor_model:
    name: extractor_model
    description: Model identifier used for extraction, e.g. claude-sonnet-4-6.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    domain_of:
    - ExtractionMetadata
    range: string
    required: true
  extractor_version:
    name: extractor_version
    description: Version string of the extraction prompt or pipeline.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    domain_of:
    - ExtractionMetadata
    range: string
    required: true
  source_text_hash:
    name: source_text_hash
    description: SHA-256 of the paper text processed, for reproducibility checking.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    domain_of:
    - ExtractionMetadata
    range: string
  extraction_date:
    name: extraction_date
    description: ISO-8601 date of extraction.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    domain_of:
    - ExtractionMetadata
    range: date
  repaired_by:
    name: repaired_by
    description: 'Identifier of the repair pass that edited this record after extraction,
      absent on a record no pass touched. Stamped because a repaired record is not
      the record the extractor produced, and treating the two as one makes a comparison
      between them meaningless.

      Declared because the pass stamps it either way: undeclared, the honest label
      was itself a validation finding, and it was the single most common thing repair
      "introduced" -- 12 of 13 across fifteen records. It was declared once by hand,
      in the generated projection rather than here, so the next `schema.generate`
      removed it again.'
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    domain_of:
    - ExtractionMetadata
    range: string
  paper_sections:
    name: paper_sections
    description: Complete deterministic index of all paper sections and subsections,
      including character spans.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    domain_of:
    - ExtractionMetadata
    range: PaperSection
    multivalued: true
    inlined_as_list: true

Induced¶

name: ExtractionMetadata
description: Pipeline provenance stamped onto the extraction record at generation
  time.
from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction
rank: 1000
attributes:
  extractor_model:
    name: extractor_model
    description: Model identifier used for extraction, e.g. claude-sonnet-4-6.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    owner: ExtractionMetadata
    domain_of:
    - ExtractionMetadata
    range: string
    required: true
  extractor_version:
    name: extractor_version
    description: Version string of the extraction prompt or pipeline.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    owner: ExtractionMetadata
    domain_of:
    - ExtractionMetadata
    range: string
    required: true
  source_text_hash:
    name: source_text_hash
    description: SHA-256 of the paper text processed, for reproducibility checking.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    owner: ExtractionMetadata
    domain_of:
    - ExtractionMetadata
    range: string
  extraction_date:
    name: extraction_date
    description: ISO-8601 date of extraction.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    owner: ExtractionMetadata
    domain_of:
    - ExtractionMetadata
    range: date
  repaired_by:
    name: repaired_by
    description: 'Identifier of the repair pass that edited this record after extraction,
      absent on a record no pass touched. Stamped because a repaired record is not
      the record the extractor produced, and treating the two as one makes a comparison
      between them meaningless.

      Declared because the pass stamps it either way: undeclared, the honest label
      was itself a validation finding, and it was the single most common thing repair
      "introduced" -- 12 of 13 across fifteen records. It was declared once by hand,
      in the generated projection rather than here, so the next `schema.generate`
      removed it again.'
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    owner: ExtractionMetadata
    domain_of:
    - ExtractionMetadata
    range: string
  paper_sections:
    name: paper_sections
    description: Complete deterministic index of all paper sections and subsections,
      including character spans.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-extraction/study
    rank: 1000
    owner: ExtractionMetadata
    domain_of:
    - ExtractionMetadata
    range: PaperSection
    multivalued: true
    inlined: true
    inlined_as_list: true