OAI-PMH (Open Archive Initiative – Protocol for Metadata Harvesting) is a client/server architecture protocol specification that facilitates the diffusion of metadata e.g.:
- Resource description of the resource (title, author, date of publishing, publisher, etc.)
- Resource location on the Internet (indicated by the URL)
The client sends a request to the server to which the response is an XML flow. The harvesting is done on non-qualified Dublin Core elements, to which additional elements (e.g. in MARC format) may optionally be attached
OAI-PMH supports two key functions:
- Metadata provision – i.e. an institution exposes its metadata (bibliographic records in Dublin Core) via an OAI=PMH server to enable them to be harvested by other organisations or Internet search engines.
- Service provision – i.e. an institution harvests metadata offered by OAI-PMH from other institutions integrating them into its own search index for use in its own user interface. The frequency of harvesting must correspond to the update frequency on the remote OAI server.
OAI-PMH is an asynchronous protocol, i.e. users do not search remote servers directly, but rather a local database that stores previously harvested metadata. Because of this, searching is faster even if a selected resource still needs to be retrieved from a remote server.
NBAs might choose to offer OAI-PMH to ensure their national bibliographies are available for indexing by general web search engines and thus can be accessed by users globally.