Scholarly metadata, deposited by thousands of our members and made openly available can act as “trust signals” for the publications. It provides information that helps others in the community to verify and assess the integrity of the work. Despite having a central responsibility in ensuring the integrity of the work that they publish, editorial teams tend not be fully aware of the value of metadata for integrity of the scholarly record. How can we change that?
Crossref was created back in 2000 by 12 forward-thinking scholarly publishers from North America and Europe, and by 2002, these members had registered 4 million DOI records. At the time of writing, we have over 23,600 members in 164 different countries. Half of our members are based in Asia, and 35% are universities or scholar-led. These members have registered over 176 million open metadata records with DOIs (as of today). What a difference 25 years makes!
In our 25th anniversary year, I thought it would be time to take a look at how we got here. And so—hold tight—we’re going to go on an adventure through space and time1, stopping every 5 years through Crossref history to check in on our members. And we’re going to see some really interesting changes over the years.
The Frankfurt Book Fair is the largest book fair in the world, and therefore a key event on our calendar. Held annually in Frankfurt, Germany, the 77th Frankfurt Book Fair (October 15–19, 2025) saw 118,000 trade visitors and 120,000 private visitors from 131 countries. The Crossref booth was located, as usual, in Hall 4.0 where all the stands with information about academic publishing can be found. Four Crossref colleagues attended the Book Fair this year, and in this blog post, you can read more about their meetings, experiences, and plans.Â
TL;DR. Metadata Manager will be retired at the end of 2025. Over the past four years, we have been developing a new helper tool to replace it, and that tool has now reached a stage of maturity that means we will be able to switch off Metadata Manager by the end of the year.
Version control is the management of changes to a document, file, or dataset. Versions of a document may include the following:
Draft
Preprint - early draft or manuscript shared by researcher in a preprint repository or dedicated channel (outside of a specific journal)
Pending publication (PP) - a manuscript which has been accepted but has not yet been published online
Advanced online publication or ahead of print (AOP) - early release of publication which publisher makes available to readers on their platform (prior to typesetting or before final published form)
Author accepted manuscript (AAM) - accepted version which has been peer reviewed but not typeset or copyedited
Version of record (VoR) - typeset, copyedited, and published version
Updated - adding supplementary data or making corrections to the file, or its retraction.
Version control is important for:
traceability (following the development of the document),
identifiability (connecting documents to decisions, contributions, contributors, and time),
clarity (distinguishing between multiple versions of documents, and identifying the latest version),
reduced duplication (removing out-of-date versions), and
reduced errors (clearly indicating to readers which is the current version).
Publication stages and DOIs
How do I decide if I should assign a DOI to a work, and at what stage? This table sets out seven publication stages of a research object (a publication such as a journal article, book, or dataset). A work may not go through all of these seven stages, so you only need to consider the stages relevant to your publication.
Publication stage
Eligible for a DOI?
Which DOI?
1 Draft
No DOI for draft item
n/a
2 Preprint
Yes
DOI A
3 Pending publication (PP)
Yes
DOI B
4 Advanced online publication/ahead of print (AOP)
Yes
DOI B
5 Author accepted manuscript (AAM)
Yes
DOI B
6 Version of record (VoR)
Yes
DOI B
7 Updated
Yes
DOI C
A DOI should not be assigned to a draft (unpublished) work.
A preprint should have its own DOI (DOI A).
Accepted versions (including PP, AOP, AAM, and VoR) should have a separate DOI (DOI B). Establish a relationship between DOI B and DOI A to show the connection between them, such as DOI B “hasPreprint” DOI A.
In the case of a significant change to the published version, a notice should be published explaining the correction/update/retraction. The updated version should have a new DOI (DOI C). Updates should only be deposited for changes that are likely to affect the interpretation or crediting of the work (editorially significant changes), and instead of simply asserting a relationship, these should be recorded as updates. See the following section for more information on updates.
Best practices for handling retractions and other post-publication updates
Research can undergo changes after it is published for various reasons. For example, it may be withdrawn, corrected, or retracted. It’s important that these changes are accurately reflected in the scholarly record, so that readers know how to find the most up-to-date work, as well as what research can be relied upon and cited.
When an editorially significant update is made to a document, you should not modify the original document, but instead issue a separate document (such as a correction or retraction notice) which explains the change. This separate document will have a different DOI and different metadata from the document that it updates. This process is complementary to versioning.
The metadata for the update should include a link to the item being updated, as well as information on the type of update, as part of the Crossmark section of the metadata:
A full example of an XML file following best practice can be found here. If you are not comfortable editing XML, you can also register Crossmark metadata using our Web Deposit Form.
Note that you don’t need to use all aspects of Crossmark to register updates. Learn more about the different ways of registering updates in our documentation.
You should also reflect the status of the work in the original DOI’s metadata record by adding “RETRACTED:” in front of the article title. We recommend doing the same for the title listed on the item’s landing page. You may also want to replace the abstract of the work with a retraction statement in both the metadata and on your website or publishing platform.
Finally, if you participate in the Similarity Check service, you should remove the full-text URL from the item’s metadata. To get a retracted work to be removed from the Similarity Check text comparison database, get in touch with Turnitin at tiisupport@turnitin.com.