TheOpen Archives InitiativeProtocol for Metadata Harvesting (OAI-PMH) is a protocol developed forharvestingmetadata descriptions of records in an archive so that services can be built using metadata from many archives. Animplementation of OAI-PMH must support representing metadata inDublin Core, but may also support additional representations.[1][2]
The protocol is usually just referred to as the OAI Protocol.
OAI-PMH usesXML overHTTP. Version 2.0 of the protocol was released in 2002; the document was last updated in 2015. It has aCreative Commons license BY-SA.
In the late 1990s,Herbert Van de Sompel (Ghent University) was working with researchers and librarians atLos Alamos National Laboratory (US) and called a meeting to address difficulties related tointeroperability issues ofe-print servers anddigital repositories. The meeting was held inSanta Fe, New Mexico, in October 1999.[3] A key development from the meeting was the definition of an interface that permitted e-print servers to exposemetadata for the papers it held in a structured fashion so other repositories could identify and copy papers of interest with each other. This interface/protocol was named the "Santa Fe Convention".[1][2][4]
Several workshops were held in 2000 at the ACM Digital Libraries conference,[5] at the 1st ACM/IEEE-CS joint conference on Digital libraries[6][7] and elsewhere to share the ideas from the Santa Fe Convention.[8] It was discovered at the workshops that the problems faced by the e-print community were also shared by libraries, museums, journal publishers, and others who needed to share distributed resources. To address these needs, theCoalition for Networked Information[9] and theDigital Library Federation[10] provided funding to establish anOpen Archives Initiative (OAI) secretariat managed by Herbert Van de Sompel and Carl Lagoze. The OAI held a meeting atCornell University (Ithaca, New York) in September 2000 aimed to improve the interface developed at the Santa Fe Convention.[11] The specifications were refined over e-mail.
OAI-PMH version 1.0 was introduced to the public in January 2001 at a workshop inWashington D.C.,[12] and another in February inBerlin, Germany.[13] Subsequent modifications to theXML standard by theW3C required making minor modifications to OAI-PMH resulting in version 1.1. The current version, 2.0, was released in June 2002. It contained several technical changes and enhancements and is not backward compatible.[14]
From 2001CERN, and later in collaboration withUniversity of Geneva, has organized bi-annual OAI workshops,[15] which over time have developed to cover most aspects ofopen science. Since 2021 the workshop series is named the Geneva Workshop on Innovations in Scholarly Communication, with the nick name OAI reflecting its origin.[16]
Some commercialsearch engines use OAI-PMH to acquire more resources.Google initially included support for OAI-PMH when launching sitemaps, however decided to support only the standard XMLSitemaps format in May 2008.[17] In 2004,Yahoo! acquired content fromOAIster (University of Michigan) that was obtained through metadata harvesting with OAI-PMH.Wikimedia uses an OAI-PMH repository to provide feeds ofWikipedia and related site updates for search engines and other bulk analysis/republishing endeavors.[18] Especially when dealing with thousands of files being harvested every day, OAI-PMH can help in reducing the network traffic and other resource usage by doing incremental harvesting.[19] NASA'sMercury metadata search system uses OAI-PMH to index thousands of metadata records from Global Change Master Directory (GCMD) every day.[20]
Themod_oai project is using OAI-PMH to expose content to web crawlers that is accessible fromApache Web servers.
OAI-PMH has later been applied to sharing of scientific data.[21]
OAI-PMH is based on aclient–server architecture, in which "harvesters" request information on updated records from "repositories". Requests for data can be based on a datestamp range, and can be restricted to named sets defined by the provider. Data providers are required to provideXML metadata inDublin Core format, and may also provide it in other XML formats.
A number of software systems support the OAI-PMH, includingFedora,EThOS from theBritish Library,GNU EPrints from theUniversity of Southampton,Open Journal Systems from thePublic Knowledge Project,Desire2Learn,DSpace fromMIT, HyperJournal from theUniversity of Pisa, Digibib from Digibis,MyCoRe,Koha, Primo, DigiTool, Rosetta and MetaLib fromEx Libris, ArchivalWare from PTFS, DOOR[22] from the eLab[23] in Lugano, Switzerland, panFMP from thePANGAEA data library,[24]SimpleDL from Roaring Development, and jOAI from theNational Center for Atmospheric Research.[25]
A number of large archives support the protocol includingarXiv and theCERN Document Server.