PROV

PROV
Name	PROV
Developer	World Wide Web Consortium; contributors from Oracle Corporation, IBM, Microsoft
Released	2013
Latest release	2013 (Core Recommendation); ongoing extensions
Programming language	agnostic
Platform	cross-platform
License	W3C Community and Contributor License

PROV PROV is a family of specifications defining a provenance data model and interchange formats. It provides a formal vocabulary and structures for representing the origins, derivation, and history of digital artifacts and activities produced by systems such as Apache Hadoop, GitHub, Wikidata, and scientific workflows used at institutions like CERN and NASA. The specifications enable interoperability among tools from vendors and projects including Oracle Corporation, IBM, and Microsoft and are maintained by the World Wide Web Consortium.

Overview

The PROV specifications define concepts for entities, activities, and agents to capture how digital items are created, used, and attributed. The design supports integration with technologies such as Resource Description Framework, RDF Schema, SPARQL, JSON-LD, and XML Schema, enabling provenance to be embedded in ecosystems like Mozilla, Google, Facebook, and scholarly infrastructures at CrossRef and ORCID. Use of PROV facilitates auditability, reproducibility, and accountability in domains represented by projects like Open Science Framework, BioConductor, Apache Airflow, and data platforms at Amazon Web Services.

History and Development

Work on the provenance family of specifications was driven by research communities around provenance systems at universities and labs including Massachusetts Institute of Technology, Stanford University, University of California, Berkeley, and Los Alamos National Laboratory. The effort consolidated ideas from provenance models such as scientific workflow provenance in Kepler, capture systems in Taverna, and databases at Oracle Corporation into an interoperable standard under the World Wide Web Consortium. Major milestones include publication of a core model and normative serializations, collaboration across vendors like IBM and standards bodies such as ISO, and adoption in governmental programs exemplified by initiatives at National Institutes of Health and European Commission research projects.

Data Model and Concepts

The core vocabulary distinguishes three primary kinds of construct: entity, activity, and agent, and defines relations such as wasGeneratedBy, used, wasAttributedTo, wasDerivedFrom, and wasAssociatedWith. Entities map to artifacts in systems like GitHub repositories, datasets in Zenodo, or documents in Microsoft Office; activities correspond to operations scheduled in tools like Apache Airflow or executed by services such as Kubernetes; agents represent actors including accounts at GitHub, organizations like Google, or researchers registered with ORCID. The model integrates with identification schemes like Uniform Resource Identifiers and persistent identifiers used by DOI registration agencies, and supports temporal constraints, roles, and bundles to represent provenance graphs captured in contexts such as ClinicalTrials.gov submissions and PubMed curation.

Serializations and Formats

PROV defines multiple serializations to suit diverse ecosystems: a Resource Description Framework/OWL mapping for semantic web platforms, a compact PROV-N notation suitable for human-readable exchange, an XML Schema variant for legacy Extensible Markup Language infrastructures, and a JSON representation often used with Node.js and web APIs. The RDF mapping permits integration with triple stores such as Apache Jena and Blazegraph and querying via SPARQL endpoints hosted by projects like DBpedia and Wikidata Query Service. Interoperability with JSON-LD enables linkage to linked data practices used by Google Knowledge Graph and digital repositories at Europeana.

Implementations and Tools

Numerous libraries and tools support the PROV model across languages and platforms: Java libraries from projects affiliated with Eclipse Foundation, Python packages used in Jupyter Notebook environments, and command-line utilities integrated in Linux distributions. Workflow systems and provenance capture frameworks such as Kepler, Taverna, Nextflow, and Airflow offer exporters to PROV formats; data repositories like Zenodo and institutional repositories at Harvard University and MIT can ingest or expose provenance metadata. Visualization and analysis tools integrate with graph databases such as Neo4j and analytics stacks built on Elasticsearch and Kibana.

Use Cases and Applications

PROV is applied in reproducible research workflows at facilities like CERN and in life sciences projects using BioConductor and Galaxy; it supports digital forensics workflows at agencies analogous to National Security Agency techniques and in legal e-discovery in law firms. Libraries and cultural heritage institutions including Bibliothèque nationale de France and British Library use provenance to track digitization processes; publishers and aggregators such as CrossRef and Elsevier integrate provenance to support article versioning and data citation. Large-scale data platforms at Amazon Web Services, Google Cloud Platform, and Microsoft Azure embed provenance to enable compliance programs in sectors regulated by laws like General Data Protection Regulation.

Standards and Governance

The PROV specifications are published and overseen by the World Wide Web Consortium working groups with input from academia, industry, and government agencies. Extensions and profiles have emerged from standards efforts and consortia including ISO workshops and research programs funded by entities like the European Commission and National Science Foundation. Governance encourages community contributions through public mailing lists, W3C working drafts, and liaison with projects such as Schema.org and Dublin Core to ensure interoperability across metadata standards.