profileFormat: ODPV-Profile-1.0
id: ODPV-OPENLINEAGE-1
title: ODPV to OpenLineage 1 profile
version: 0.1.0
status: stable
vocabulary:
  id: ODPV
  version: 1.1.0
  baseUri: https://opendataproducts.org/odpv-v1.1/
externalStandard:
  name: OpenLineage
  version: '1.x'
  specification: https://openlineage.io/docs/spec/object-model/
purpose: Connect declared ODPV product semantics with observed OpenLineage dataset, job, and run metadata without treating runtime observations as business or contract proof.
scope:
  includes:
    - Dataset and workflow bindings
    - Declared and observed lineage distinctions
    - Dataset quality assertion bindings
  excludes:
    - A replacement OpenLineage event schema
    - Proof of contract compliance or business value
    - Policy evaluation
mappings:
  - id: dataset
    kind: fieldBinding
    odpvTerm: Dataset
    externalReference: OpenLineage Dataset
    rationale: Both identify a discrete collection of data, but an OpenLineage Dataset is an operational identity within a namespace and name.
    evidence: https://openlineage.io/docs/spec/object-model/
    status: reviewed
  - id: workflow-job
    kind: fieldBinding
    odpvTerm: Workflow
    externalReference: OpenLineage Job
    rationale: An OpenLineage Job can implement a workflow step, but an ODPV Workflow may be broader than one executable job.
    evidence: https://openlineage.io/docs/spec/object-model/
    status: reviewed
  - id: owner
    kind: fieldBinding
    odpvTerm: Owner
    externalReference: OpenLineage Ownership Dataset and Job facets
    rationale: OpenLineage can record technical ownership for datasets and jobs; the facet does not establish all accountability captured by an ODPV Owner.
    evidence: https://openlineage.io/docs/1.50.0/spec/facets/job-facets/ownership/
    status: reviewed
  - id: uses
    kind: transformation
    odpvTerm: uses
    externalReference: OpenLineage input Dataset
    direction: odpvToExternal
    rationale: An ODPV use edge becomes an input dataset only when the source is represented by a concrete executing job and the target is an operational dataset.
    evidence: https://openlineage.io/docs/spec/object-model/
    status: reviewed
    implementationNotes: The SDK must reject a direct conversion when the ODPV source is not bound to an OpenLineage Job or the target is not bound to a Dataset.
  - id: consumes
    kind: transformation
    odpvTerm: consumes
    externalReference: OpenLineage input Dataset
    direction: odpvToExternal
    rationale: Consumption maps to a job input at an observed or declared execution boundary, not to every conceptual dependency in an ODPV graph.
    evidence: https://openlineage.io/docs/spec/object-model/
    status: reviewed
    implementationNotes: The SDK must retain whether the source came from declared graph metadata or an observed event.
  - id: declared-lineage
    kind: transformation
    odpvTerm: derivedFrom
    externalReference: OpenLineage JobEvent lineage facet
    direction: odpvToExternal
    rationale: Design-time lineage can express declared derivation, while run events express an observation during a particular execution.
    evidence: https://openlineage.io/docs/spec/facets/job-facets/lineage/
    status: reviewed
    implementationNotes: The SDK must label derived lineage as declared when emitting a JobEvent and must retain event type when importing observed lineage.
  - id: production-boundary
    kind: transformation
    odpvTerm: produces
    externalReference: OpenLineage output Dataset
    direction: odpvToExternal
    rationale: A job output represents an operational production boundary; it does not by itself establish the broader product relationship or business outcome.
    evidence: https://openlineage.io/docs/spec/object-model/
    status: reviewed
    implementationNotes: Emit outputs as observed only for RunEvent data and retain the job identity that owns the output.
  - id: quality-assertions
    kind: fieldBinding
    odpvTerm: DataQuality
    externalReference: OpenLineage DataQualityAssertionsDatasetFacet
    rationale: Assertion results are evidence about tests on an observed dataset, not a complete definition of the ODPV DataQuality concept.
    evidence: https://openlineage.io/docs/spec/facets/dataset-facets/data_quality_assertions/
    status: reviewed
  - id: quality-metrics
    kind: fieldBinding
    odpvTerm: DataQuality
    externalReference: OpenLineage DataQualityMetricsDatasetFacet
    rationale: Dataset and column metrics are operational quality evidence, not a complete ODPV DataQuality definition or guarantee.
    evidence: https://openlineage.io/docs/spec/facets/dataset-facets/data_quality_metrics/
    status: reviewed
  - id: dataset-version
    kind: fieldBinding
    odpvTerm: versionOf
    externalReference: OpenLineage DatasetVersionDatasetFacet
    rationale: A dataset-store version can support a version relationship, but not every ODPV version is a physical dataset snapshot.
    evidence: https://openlineage.io/docs/spec/object-model/
    status: reviewed