3 minute read.It’s Time: Planning for Metadata Schema Deprecation
It has been 18 (!) years since Crossref last deprecated a metadata schema. In that time, we’ve released numerous schema versions, some major updates, and some interim releases that never saw wide adoption. Now, with 27 different schemas to support, we believe it’s time to streamline and move forward.
Starting next year, we plan to begin the process of deprecating lightly-used schemas, with the understanding that this will be a multi-year effort involving careful planning and plenty of communication.
Which schema will be deprecated?
There are two types of schema used to register content metadata records: a full metadata input schema, which follows the pattern crossrefX.X.X.xsd, and resource schema, which follows the pattern doi_resourcesX.X.X.xsd. The resource schema are used to append metadata, such as references or funding data, to an existing metadata record.
I’ve categorized our schemas by usage levels to help prioritize the deprecation process:
Light usage (planned for initial deprecation):
- crossref4.3.1.xsd
- crossref4.3.2.xsd
- crossref4.8.1.xsd
- doi_resources4.3.2.xsd
- doi_resources4.3.4.xsd
- doi_resources4.3.5.xsd
- doi_resources4.4.2.xsd
Moderate usage:
- crossref4.3.3.xsd
- crossref4.3.4.xsd
- crossref4.3.5.xsd
High usage:
- crossref4.3.0.xsd
- crossref4.3.6.xsd
- crossref4.3.7.xsd
- crossref4.4.0.xsd
- crossref4.4.1.xsd
- crossref4.4.2.xsd
- doi_resources4.3.0.xsd
- doi_resources4.3.6.xsd
We currently support 5 versions of our grants-specific schema and will be working with our funder members to move to new versions of that schema over time - this will follow a different timeline and process as there are fewer schemas to navigate.
If you don’t know which version you’re currently using, now would be a good time to check. Many of our members are still using 4.3.0, the earliest supported version.
Why deprecate now?
- Supporting 27 schema is unsustainable: Each schema version we maintain adds complexity to our systems and makes it harder to implement improvements that benefit everyone.
- Existing schema need modernization. Some fundamental elements, like names and titles, need to be modeled differently to fully capture variations in language and usage patterns across different cultures and contexts. We also have too many bespoke record types. Consolidating these will create a simpler, more coherent structure. We may retain certain specialized structures for journal articles and books, but overall, simplification will benefit everyone.
Most importantly:
- Our current requirements are too minimal. For most record types, we only require a title and publication year. While this low barrier has made registration accessible, it hasn’t served metadata quality well. We know you can do better, and we’d like to ask for more to improve the richness and utility of Crossref metadata.
What happens next?
This won’t be an abrupt change. We would like to deprecate the schema flagged ‘light usage’ by the end of 2026 and will be reaching out to impacted members early next year. For other schema, we’re planning a multi-year effort with clear communication at every stage. We’ll provide ample notice before any schema is deprecated, along with migration guidance and support.
With the exception of recent changes to affiliation metadata, we’ve primarily been building on existing schema structures. This means upgrading should be straightforward for most users. As mentioned, we’ll judiciously making some breaking changes to names, titles, and requirements, and would like to consolidate schema as we move forward.
Our goal is to create a more robust, modern metadata framework that better serves the scholarly community while reducing the maintenance burden that comes with supporting decades of schema versions. Stay tuned for more details on timelines and migration paths. In the meantime, if you’re unsure which schema version you’re using, we encourage you to check your current implementation.
