For immediate release June 14, 2002


Open Archives Initiative Release Version 2.0 of the Protocol for Metadata Harvesting


Ithaca, NY & Los Alamos NM—The Open Archives Initiative is pleased to release version 2.0 of the Open Archives Initiative Protocol for Metadata Harvesting (OAI-PMH). The release of OAI-PMH v.2.0 comes after 16 months of worldwide experimentation with version 1.x of the protocol, an 8 month revision process by the OAI-tech group, and 4 months of alpha/beta testing. Thanks to this rigid testing and revision, we feel confident to release the OAI-PMH version 2.0 as a stable specification.

Summary of changes in version 2.0

  • Clarifications and Specification Cleanups.
    • Core Protocol versus Extensible Notions. A clear boundary has been established in between the core OAI-PMH protocol and ancillary concepts, by separating the protocol specification from a newly introduced Implementation Guidelines.
    • Clarified Data Model. The protocol now builds on a well-defined “resource – item – metadata-record” data model, correcting a level of ambiguity that existed with this respect in v.1.x.
    • Dublin Core Metadata Definition. In a successful and trend-setting collaboration with the Dublin Core Metadata Initiative, an XML Schema for unqualified Dublin Core has been created, which is hosted by the DCMI and used in the delivery of metadata in the mandatory DC format in the OAI-PMH.
    • Single XML Schema for OAI-PMH Responses. OAI-PMH now defines a single XML Schema to validate responses to all OAI-PMH requests.
    • Language Cleanup. The OAI-PMH now uses the notions must, must not, may, should, etc. as in RFC2119, allowing for a better understanding of protocol conformance.
  • Error Corrections.
    • Uniform Date and Time Encoding. Dates and times are now uniformly encoded using ISO8601 and are expressed in UTC throughout the protocol.
    • Flow Control Idempotency. In order to address known problems with harvesting large datasets, the new protocol clearly defines and mandates the idempotent nature of resumptionTokens, providing deterministic behavior in the face of data transmission errors.
  • New Functionality.
    • Multiple Time Granularity. In response to requests originating mainly from the union-catalogue realm, the OAI-PMH now supports harvesting at different levels of time-granularity: support of day granularity is mandatory, while support for seconds granularity is optional.
    • Error Reporting. Version 2 introduces a new error reporting mechanism, clearly separating between errors occurring at the HTTP transport layer used by the OAI-PMH and errors at the OAI-PMH level.
    • Enhanced Identify Semantics. The response to the Identify request is more expressive than in Version 1.x, providing more information to harvesters and supporting multiple harvesting strategies.
    • Item Set Membership. For repositories that support set-structures, it is now mandatory to list set membership of items in the responses to the GetRecord, ListRecords and ListIdentifiers requests.
  • Related Changes and Activities.
    • Implementation Guidelines. Implementation Guidelines documents are provided to support interpretation of the protocol document by implementers of repositories, harvesters, aggregators, mirrors, caches, etc.
    • Friends Schema. The usage of a “friends” XML Schema at the level of the Identify-response is recommended to help establish a dynamic approach to the process of discovering repositories.
    • Provenance Schema. The usage of a "provenance" XML Schema is recommended at the level of the "about" container attached to metadata records, in order to describe the provenance of exposed metadata, and support deduplication of metadata harvested from multiple repositories.
    • Branding Schema: An XML Schema for usage at the repository or set-level is provided to facilitate branding and rendering of harvested metadata in third party services.
    • MARC XML: The usage of the MARCXML Schema released by the Library of Congress (June 2002) is recommended over the usage of the oai_marc Schema that was introduced with the release of version 1.0 of the OAI-PMH.
    • Identifiers: The specification for the optional oai-identifier scheme for uniquely identifying items in a repository now relies on dns-names instead of registered repository-identifiers to achieve global uniqueness.

Strategy for transition from OAI-PMH 1.1 to 2.0

The OAI registration service for data providers is ready to accept registration of 2.0 conformant repositories at  On September 1st 2002, registered sites that have not upgraded to version 2.0 will be sent a reminder that they must upgrade. As of this date, attempts to register with the OAI registration service must conform to OAI-PMH version 2.0.  On December 1st 2002, repositories that have not upgraded to 2.0 will be removed from the OAI registry.

Essential links to information about the release

Links to version 2.0 tools.


This release is a good opportunity to acknowledge institutions and individuals who have significantly contributed to the remarkable endeavor that started with the first Open Archives Meeting in 1999 in Santa Fe and that led to this official release of the OAI-PMH v.2.0 protocol.

For long-term financial support of OAI activities and development:

  • Coalition for Networked Information
  • Digital Library Federation
  • National Science Foundation

For financial support and support-in-kind of various OAI activities:

  • Cornell University
  • Council on Library and Information Resources
  • Dublin Core Metadata Initiative
  • Library of Congress
  • Los Alamos National Laboratory
  • Andrew W. Mellon Foundation

For participation in the development of the OAI-PMH and OAI support activities:


Carl Lagoze; & Herbert Van de Sompel;  The OAI Executive.

  June 14, 2002

get in touch with the Open Archives initiative by contacting

the Open Archives home page