DataTerminology.comNEW
Also known as: PROV Graph
A graph model that captures entities, activities, and agents to describe how data was produced and transformed.
DESCRIPTION
A provenance graph represents derivations and dependencies among datasets, processes, and actors over time. It enables auditability, reproducibility, and explainability by linking outputs to their inputs and transformations. The W3C PROV family standardizes concepts and relations for interoperability. Provenance graphs can be stored as RDF triples and queried with SPARQL. They support impact analysis and regulatory reporting. Capturing provenance requires instrumenting pipelines and tools. Aligning business and technical lineage enhances trust and governance.
ALTERNATIVE DEFINITIONS
W3C PROV Overview
PROV is a set of specifications to represent and interchange provenance, i.e., information about entities, activities, and people involved in producing a piece of data.
Formalizes provenance and its core elements for interchange.
W3C PROV-O
The PROV Ontology (PROV-O) provides classes, properties, and restrictions to model provenance information for interoperable exchange.
Defines an OWL ontology for provenance representation.
RELATED TERMS
W3C PROV PROV-O Knowledge Graph RDF Knowledge Representation Workflow Directed Acyclic Graph Metadata Management
COMPETING TERMS
Lineage Transaction Audit Data
RELATIONS
- Broader: Graph
KNOWLEDGE AREAS
Metadata Management Semantic & Knowledge Modeling Data Governance Data Management Data Science
TARGET AUDIENCE
Data Steward Data Governance Manager Data Architect Data Engineer Data Scientist Chief Data Officer
EXTERNAL SOURCES
- Wikipedia - PROV
- W3C - Overview of PROV specifications
- W3C - Provenance ontology (PROV-O)
- OpenLineage - Open lineage standard and spec
TOP QUESTIONS
Why is a Provenance Graph important?
It provides traceability from outputs back to inputs and processes, enabling auditability, reproducibility, and impact analysis across data pipelines.
Category: Relevancy
How does PROV relate to a Provenance Graph?
PROV defines standardized entities, activities, and agents, along with relations, making provenance graphs interoperable across tools and organizations.
Category: Definition
What’s a common mistake capturing provenance?
Collecting only coarse-grained pipeline steps without versioned inputs, parameters, or environment details limits reproducibility and trust.
Category: Mistakes
What is the risk of missing provenance?
Without provenance, compliance reporting, root-cause analysis, and model explainability become difficult or impossible, increasing regulatory and operational risk.
Category: Risks
How does a Provenance Graph differ from an audit log?
An audit log records events, while provenance models causal relationships among entities, activities, and agents to explain derivations.
Category: Alternatives
What technologies store provenance graphs?
RDF triple stores, graph databases, or JSON-LD stores are common. SPARQL enables rich queries over PROV-O representations.
Category: Example
How is provenance used in machine learning?
It tracks datasets, features, hyperparameters, and code versions, enabling experiment reproducibility and model governance.
Category: Example
What’s a hot topic around provenance?
Integrating operational lineage with business context and policy controls to build end-to-end, queryable knowledge graphs.
Category: Topic
Where should I start with provenance?
Adopt a standard like PROV or OpenLineage, instrument pipelines to emit events, and centralize storage with queryable metadata.
Category: Best practices