1. Open Archive Initiative – Protocol for metadata Harvesting (OAI-PMH) Surinder Kumar Technical Director NIC, New Delhi [email_address] , 011-24305503
1st Meeting of the OAI October 21-22 1999, Santa Fe, New Mexico D-Lib article in Feb 2000: http://www.dlib.org/dlib/february00/vandesompel-oai/02vandesompel-oai.html
The service provider will use a “harvester” to issue OAI requests in order to collect metadata from a data provider’s repository. Example services: Anything you can do with metadata. Cross search: aggregate resource discovery metadata from many repositories (BUT OAI-PMH is not a search protocol per se ). Metadata enhancement/transformation: merge records, augment metadata (e.g. add implicit information about encoding schemes), change binding. (Diane Hillman et al “Improving metadata quality : recombination and augmentation” (NSDL) http://metamanagement.comm.nsdl.org/Metadata_Augmentation--DC2004.html
Will focus on those which are highlighted.
These are examples using HTTP GET, also possible to use POST
By insisting the simple DC is used the OAI aims to provide a base level of interoperability.
Selective harvesting is not the same as searching, sets have to be pre-defined by the data provider. You can’t assume that an OAI data provider will provide sets.
OAI is not a query protocol.
STARGATE (CDLR at Strathclyde university) http://cdlr.strath.ac.uk/stargate/ is investigating use of OAI static repositories
None of the issues are insurmountable if you’re not over-ambitious to start with and think through potential problems. NSDL metadata recombination work is an example.
Commercial uptake: in work with a scientific publisher (Inderscience) through JISC PALS project, have found that CSA, CrossRef, Elsevier etc. have refused to use OAI-PMH preferring their own harvesting techniques. In part perhaps due to not understanding how flexible OAI can be (e.g can support any metadata) In part because can do this simply without OAI-PMH (e.g. web crawling) In part reflects balance of power between service provider and (small) data provider (DP has to play by SP rules; DP has to support many harvesting specs, SP get to control what they use) PERX project have produced “Marketing with metadata” aimed at encouraging publishers to use metadata interoperability standards including OAI-PMH Support for rich metadata reflects variety and lack of consensus for rich metadata
Location: no metadata element is mandatory, therefore can’t rely on identifier/location being available. However if you’re dealing with resource discovery metadata for resources available online, it normally is.