Crossref operates an OAI-PMH service for the distribution of metadata to subscribers (Enhanced CMS). This system is based on the OAI-PMH version 2 repository framework and implements the interface as documented at http://www.openarchives.org/OAI/openarchivesprotocol.html.
We allow public access to two OAI verbs, ListSets and ListIdentifiers, which allow for discovery of coverage information. Access to OAI verbs GetRecord and ListRecords require a subscription.
The OAI-PMH service is open to Enhanced CMS subscribers only. Subscribers may provide 2 IP addresses or CIDR ranges for authentication. Token-based access is also available.
We support selective harvesting according to sets defined by the hierarchy of publisher and title. Setspecs are formatted as follows:
- content type:prefix:pubID (ex: J:10.1002:4 = Journal content by the publisher Wiley, journal title Applied Organometallic Chemistry)
- content type:prefix (ex: J:10.1002, journals owned by publisher Wiley)
The from and until dates in a request capture when a record was deposited or updated, not the published date of the item. This means a request for records from yesterday through today will return all records added or changed between then and now, regardless of the publication dates included in the records.
Set content types are:
- J for journals
- B for books, conference proceedings, dissertations, reports, and datasets
- S for series
The default set for both ListIdentifiers and ListRecords is J (journals). A set (B for books or conference proceedings, S for series) must be specified to retrieve non-journal data.
With the ListSets request the set parameter is optional. Leaving off the set parameter will return only journal data which includes a list of publishers, their journal titles, and each year of publication for which we have metadata records.
With the ListIdentifiers request the set, from, and until parameters are optional. The from and until parameters are used to specify dates when the DOIs were registered with Crossref and not the publication date.
Request a list of DOIs registered since 2010-08-11:
Request all journal sets:
Request all sets with content type 'B':
Request records for title '98765' with prefix 10.1234 registered or updated on 2017-07-06:
Best practice for performance
We allow 3 concurrent initial OAI-PMH requests per user. There is no concurrency limit for follow-on requests (requests made with a resumption token). Due to the size of the repository it is highly discouraged to perform a ListRecords action for the entire collection.
The best possible performance is had by requesting one publication's day's changes, eg
If you are harvesting a large amount of data and run up against our 3 concurrent initial request limitation, it is recommended that you request data by prefix for a short time frame (days to a week). For example, this request will give you all journal records owned by prefix 10.1234 registered or updated between 2017-07-06 and 2017-07-09 :
Many OAI requests are too big to be retrieved in a single transaction. If a given response contains a resumption token, the user must make an additional request to retrieve the rest of the data. Resumption tokens remain viable for 48 hours.
The resumption token includes an expiration date of 48 hours:
The token should be appended to the end of the next request:
Initial data loads
As mentioned above, the size of the Crossref repository precludes using one OAI request to retrieve all the available data. Upon request we can produce an archive of all the data available to a given subscriber which can then be downloaded via FTP. We limit archive requests to one per year.
A sample application for harvesting Crossref OAI data is available below (oaipmhRequest.zip). Sample ListIdentifiers, ListSets, and ListIdentifiers responses are also available (oaipmhSamples.zip).