The problem with duplicate DOIs and how you can help

When a DOI is registered for an item, it should remain the one persistent identifier for that item - forever. Members should not register another DOI for an item that already has one. Having more than one DOI for a work is a problem for everyone. It means that the work is less likely to be matched when someone cites it, it splits the citation record, and it undermines the purpose of registering persistent identifiers. It also makes a mess of that all-important metadata.

However, we do know that there are duplicate DOIs out there. There are many possible reasons for this. It might be that after we transferred a title between two members, the acquiring member’s automated processes kicked in and accidentally registered new DOIs for existing content. It might be that the member didn’t realize that there were existing DOIs for the content they had acquired. Or it might be that the member was confused, and didn’t understand the damaging implications of having multiple DOIs.

There are no easy or short-term solutions to fully address the problem of duplicate DOIs, but everyone has a part to play in the solution.

What Crossref is doing

1. Trying to stop duplicate DOIs being registered after a title transfer between Crossref members

Registering a DOI for content that already has a DOI is in contravention of the Crossref member terms. When working on title transfers between members we reiterate this, and provide a list of the existing DOIs for the titles being transferred. We hope that this will help to reduce the problem going forward at the source.

2. Making it clearer to members that they’ve registered a duplicate DOI

We currently don’t block members from registering content that seemingly already has a DOI, as determining a truly duplicate DOI requires human review by someone who understands the content. If we automated blocks of potentially duplicated content, it would likely prevent registration of things like Book Reviews, Letters, and Errata. This content often shares a single page and may have no author, thus resulting in limited and matching metadata between truly distinct content.

However, we send all members a conflict report each month, which highlights where we believe that they may have a duplicate DOI. We know that members find this report difficult to use and understand, and we have plans to improve the report.

Currently, members can use this report to confirm if the conflict genuinely is a duplicate DOI (and set up an alias between the two DOIs) or confirm if they are indeed two separate citable objects with similar metadata.

When a member sets up an alias, one of the two DOIs is set as the Primary. This is the DOI whose metadata will be kept up to date going forward. The other DOI is set as the alias. Anyone following the aliased DOI will resolve to the resource resolution URL of the Primary DOI.

3. Getting a better idea of the scale of the problem

Our colleague Dominika did some research into this problem back in 2020, and in her sample of 524,496 DOIs, 0.8% were flagged as having non-aliased duplicates. Our R&D team is undertaking a short-term project to research this further, to get an idea of the scale of both non-aliased and aliased DOIs, and an idea of whether the problem is reducing or not. We’ll be updating this post with what we discover.

4. Making it clearer for our metadata users when a DOI is a possible duplicate or a confirmed duplicate

We don’t delete duplicate DOIs, as they are often already displayed and have potentially been used to cite work. With a confirmed duplicate, we can set up an alias, but the fact that a DOI is the Primary or the alias is not currently present in our REST APIs. However, we’re working on plans to add this. You can watch the development of these plans here.

We’re also considering adding the fact that there is an unresolved conflict between two DOIs into our metadata, but this conversation is at a very early stage.

5. Proactively contacting members who are still registering duplicate DOIs

Where we are advised that members seem to be registering duplicate DOIs, we contact them to explain the implications and that it is in contravention of the membership terms. We have not been doing this proactively, but we’re hoping to improve our reporting on duplicate DOIs to allow us to do this more systematically in the future. We’re also considering making this information more public too.

How Crossref members can help

If you acquire a title with existing DOIs, make sure that you continue to use and display the existing DOIs. Our depositor reports are a good starting point for determining what has already been registered for a specific title. We’ll always provide you with a link to the depositor report for the relevant title when we perform a title transfer. It’s also key to amend any automated processes you may have so that you don’t accidentally register duplicate DOIs for titles that you acquire.

We’re hoping that the combination of these actions will reduce new duplicate DOIs from being registered, and make it clearer where duplicate DOIs exist. If you have any suggestions for how we can help further, do let us know in the comments.