DataCite Metadata Normalization Changes

What’s changing

When metadata is submitted with DataCite to register or update a DOI, DataCite may apply isolated changes (or "normalizations") to some submitted metadata fields, particularly when converting submitted XML metadata to JSON for appearance in REST API responses. These behaviors have caused confusion for some users and are no longer aligned with our approach to metadata submission.

We’re planning to phase out the normalizations outlined below. These changes will take effect in Q1 2026.

What do I need to do?

You don’t need to change anything about how you are submitting metadata, and all valid API calls to the MDS API and REST API will remain valid. The appearance of certain metadata may change slightly in REST API responses for new and updated DOIs according to the changes described below.

What normalizations will be phased out?

The following normalizations will be removed:

Affected submission methodsDeprecated behavior
MDS API/XMLIf there is not already a Date with dateType "Issued", one is inserted into the REST API response using the submitted PublicationYear value. XML metadata is not modified.
MDS API/XMLCreator and Contributor name, givenName, and familyName metadata is interpreted to normalize the name value and populate the nameType value in the REST API response. XML metadata is not modified.
MDS API/XML
REST API
When a subject can be matched to a term from the OECD’s Fields of Science (FOS) vocabulary, a subject with the FOS term is inserted into the REST API response. XML metadata is not modified.

Are there other normalizations still in place?

Yes. The following normalizations will remain in place pending ongoing assessment:


Affected submission methodsNormalization behavior
MDS API/XML
REST API
Three-letter language codes (ISO 639-3) are updated to two-letter codes (ISO 639-1) when an exact match is available. Country codes are also removed. For more information, see: Why do the language codes in the metadata change?
MDS API/XML
REST API
If the Rights name, rightsURI, or rightsIdentifier matches a license in SPDX, we add the corresponding rightsUri, rightsIdentifier, rightsIdentifierScheme, and schemeUri values

👍

We’d love your feedback on either maintaining or removing these normalizations. Contact [email protected] and let us know what you think!