Namespace doc: is https://ns.flur.ee/doc#. Structure elements additionally use DoCO (doco:, http://purl.org/spar/doco/), the pattern ontology it extends (po:, http://www.essepuntato.it/2008/12/pattern#) and NIF (nif:, http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#). The structure terms are those of the parsing engine and are documented in full in fluree-doc-parse ; the terms below are what the ingest adds.
Class Meaning
doc:ChunkA retrieval unit cut from a document’s structure.
doc:SourceDocumentThe file a graph was extracted from; the doc:sourceDocument target.
doc:MentionA span of a chunk naming an entity. Also typed nif:RFC5147String.
doc:RelationOne relation the language model reported, reified with its evidence and verdict.
doc:EntityAn entity extraction minted because no --entities source knew it. Known entities keep their own IRI and get no node here.
Property Range Meaning
doc:chunkIndexinteger Position in the document.
doc:textstring The chunk’s text, whitespace collapsed.
doc:headerPathstring Section titles above it, outermost first, joined with /. Absent when there are none.
doc:sourceElementIRI, repeated The elements it was built from, in order.
doc:sourceDocumentIRI The document. Also on every element.
doc:embedding@vectorIts embedding, when produced.
Property Range Meaning
doc:fileNamestring The file’s name.
doc:relativePathstring Path relative to the ingested directory, /-separated.
doc:sha256string Hex digest of the bytes.
doc:mediaTypestring The media type inferred from the extension.
doc:byteSizeinteger Size in bytes.
doc:pageCountinteger Pages, for sources that have them; 0 otherwise.
doc:escalatedCropsinteger Crops a vision model read for this extraction.
doc:parserRevisionstring The fluree-doc-parse revision that produced the structure graph.
doc:chunkCountinteger Chunks written.
doc:chunkingstring min/max characters the chunks were cut with.
doc:embeddingModelstring The embedding model, when one ran.
doc:embeddingDimensionsinteger Its vector width.
doc:ingestedAtxsd:dateTimeWhen this extraction was written.
doc:extractionModelstring The language model that extracted, when one ran.
doc:extractionFingerprintstring Hash of the ontology, entity sources, model, guidance and relation mode.
doc:mentionCountinteger Mentions written.
doc:entityCountinteger Distinct entities mentioned, known and new.
doc:relationCountinteger Relations written, every verdict included.
Property Range Meaning
nif:beginIndex, nif:endIndexinteger Character offsets into the chunk’s doc:text.
nif:anchorOfstring The text as written.
nif:referenceContextIRI The chunk.
nif:entityIRI The entity, under the IRI its source gave it.
doc:sourceElementIRI The structure element the span sits in.
doc:extractedBystring gazetteer or llm.
doc:sourceDocumentIRI The document.
Property Range Meaning
rdf:subject, rdf:predicateIRI The statement.
rdf:objectIRI or literal The statement’s object: an entity, or the literal the model gave when the object was a value or named no entity in the chunk.
rdfs:labelstring `subject
doc:excerptstring The text supporting it, as the model quoted it.
doc:verdictstring valid, repaired or rejected.
doc:assertedboolean Whether a direct edge was written for it.
doc:originalPredicate, doc:repairNotestring On a repaired relation: what the model wrote, and how it was resolved.
doc:rejectionReasonstring On a rejected relation.
doc:sourceChunk, doc:sourceDocumentIRI Where it was found.
Property Range Meaning
schema:namestring The canonical name.
skos:altLabelstring, repeated Other surface forms seen.
doc:nerLabelstring Coarse label: PERSON, ORG, GPE, LOC, FAC, EVENT, PRODUCT, WORK_OF_ART, LANGUAGE, MISC, CONCEPT.
doc:offModelboolean Present and true when the model’s class for the entity is not in the ontology; the node carries no class.
ontology datatype properties literal Attributes the model stated and the ontology admits.
doco:Document, doco:BodyMatter, doco:Section, doco:SectionTitle, doco:Title, doco:Paragraph, doco:ListItem, doco:Caption, doco:Table, doc:TableCell (doc:cellValue, doc:rowHeader, doc:columnHeader, doc:rowIndex, doc:columnIndex), po:contains, nif:isString, nif:beginIndex, nif:endIndex, doc:sectionLevel, doc:pageIndex, doc:bbox, doc:pages, doc:evidence, doc:provenance.