Skip to main content

PubMed (RDF/Pydantic Ontology Models)

What it is​

  • A small set of Pydantic models representing a PubMed RDF ontology.
  • Each model can generate an rdflib.Graph of RDF triples for the instance (and nested related instances) via rdf().

Public API​

Base class​

  • class RDFEntity(BaseModel)
    • Purpose: common RDF/URI behavior for all ontology entities.
    • Public attributes:
      • _namespace: ClassVar[str] (default: http://example.org/instance/) used to auto-generate instance URIs.
      • _uri: str instance URI (auto-generated if not provided).
    • Public methods:
      • set_namespace(namespace: str) -> None: set the namespace used for new auto-generated URIs.
      • rdf(subject_uri: str | None = None) -> rdflib.Graph: produce RDF triples for the instance.
        • Adds rdf:type using _class_uri (if present).
        • Adds property triples based on _property_uris (if present).
        • Recursively includes RDF for related objects that implement .rdf().

Ontology entity classes​

All below inherit from RDFEntity and are Pydantic models with extra='forbid'.

  • class PubMedPaperSummary(RDFEntity)

    • Purpose: metadata summary returned by the PubMed API for a single paper.
    • Data fields (optional):
      • authorLiteral: str
      • doi: str
      • downloadUrl: Any
      • journalTitleLiteral: str
      • pages: str
      • pmcid: str
      • publicationDate: datetime.date
      • pubmedIdentifier: str
      • sortPublicationDate: datetime.datetime
      • title: str
      • url: Any
    • Object fields (optional):
      • aboutJournal: Journal
      • aboutJournalIssue: JournalIssue
      • hasAuthorshipRole: AuthorshipRole
      • summarizes: PubMedPaper

  • class PubMedPaper(RDFEntity)

    • Purpose: canonical description of a PubMed article.
    • Object fields (optional):
      • hasAbstract: LiteralContent
      • hasKeyword: LiteralContent
      • hasMeshDescriptor: LiteralContent
      • publishedIn: JournalIssue

  • class LiteralContent(RDFEntity)

    • Purpose: holds literal text values.
    • Data fields (optional):
      • literalValue: str

  • class Journal(RDFEntity)

    • Purpose: serial publication venue.
    • Data fields (optional):
      • issn: str

  • class JournalIssue(RDFEntity)

    • Purpose: denotes a specific journal issue.
    • Data fields (optional):
      • issueLabel: str
      • volume: str
    • Object fields (optional):
      • issueOf: Journal

  • class Author(RDFEntity)

    • Purpose: a human who bears an authorship role for a PubMed paper.
    • No declared fields/properties beyond RDFEntity internals.

  • class AuthorshipRole(RDFEntity)

    • Purpose: a role held by a person when they author a PubMed paper.
    • Object fields (optional):
      • roleHeldBy: Author

Configuration/Dependencies​

  • Python dependencies:
    • pydantic (models, validation)
    • rdflib (Graph, URIRef, Literal, RDF)
  • Model configuration:
    • arbitrary_types_allowed=True
    • extra='forbid' (unknown fields raise validation errors)
  • URIs:
    • If _uri is not provided, a URI is auto-generated as f"{_namespace}{uuid.uuid4()}".
    • The RDF predicate URIs and class URIs are hard-coded per class via _property_uris and _class_uri.

Usage​

from rdflib import Graph
from naas_abi_marketplace.applications.pubmed.ontologies.PubMed import (
    RDFEntity, PubMedPaperSummary, PubMedPaper, LiteralContent, Journal, JournalIssue
)

# Optional: set base namespace for generated instance URIs
RDFEntity.set_namespace("http://my.example/instance/")

journal = Journal(issn="1234-5678")
issue = JournalIssue(volume="42", issueLabel="1", issueOf=journal)

paper = PubMedPaper(
    hasAbstract=LiteralContent(literalValue="Abstract text."),
    publishedIn=issue
)

summary = PubMedPaperSummary(
    title="A paper",
    doi="10.0000/example",
    pubmedIdentifier="123456",
    aboutJournal=journal,
    aboutJournalIssue=issue,
    summarizes=paper
)

g: Graph = summary.rdf()
print(len(g))          # number of triples
print(g.serialize())   # default serialization (format chosen by rdflib)

Caveats​

  • rdf() treats non-model values as RDF literals; object values are linked by their ._uri and their own triples are included.
  • Lists are supported in rdf() generation, but the ontology fields in this file are not typed as lists; passing lists may still work at runtime but must satisfy Pydantic validation (or use Any-typed fields).
  • downloadUrl and url are typed as Any and will be emitted as RDF literals unless you supply an object that implements .rdf() (in which case it will be treated as a related RDF entity).