Define and implement workflow for updating Datacite DOIs

[JessyBarrette]

When the version element is updated by a user, it needs to be updated in DataCite as well as the Hakai Catalogue… How to do this? Can we dynamically update Datacite records automatically?

[fostermh]

• Can a 'findable' datacite entry be updated?
• Can we update the datacite record on publishing a metadata entry form record?

[Br-Johnson]

A findable datacite DOI can indeed be updated using the metadata intake form, which as we discovered is fraught with danger because the current DOI metadata and any manual edits are overwritten.

[Br-Johnson]

Here's a draft of the workflow for DOI creation that Tim worked up:
https://lucid.app/lucidchart/80acd67e-da43-44cc-8dc7-a47e7301913f/edit?viewport_loc=-2277%2C-421%2C3002%2C1507%2C0_0&invitationId=inv_226cbe12-5469-47c3-8e5b-259e026edda8

[fostermh]

hmm seems like the 'update a published record' workflow needs some adjustment then.

Should an update to a published record require another review?
If updating a doi can we pull down the existing record first, compare changes, and ask the user if it's ok to overwrite?
Will an update to a doi never be ok because we do not populate fields correctly?

[Br-Johnson]

Good questions.

Just talked about this with the DM crew and 1) no we don't think updating a published record requires another review.
2) That might be a bit overkill unless we want to implement metadata versioning, but not clear why we'd need that if we're not reviewing the diffs. Unless you're talking about pulling down the DOI metadata and showing the diff to the user and asking if it's OK they overwrite? Might be too much...? Ideally the metadata intake form is the signle source of truth for DOI metdata.
3) Currently the mapping isn't perfect, but Austen has a PR in review and if we can get it good then it would be best case to rely on that instead of manual changes and review.

Another question we have is whether or not a manually entered DOI can be updated. I can test this.

[Br-Johnson]

A manually entered DOI can not be updated using the metadata intake tool . So, all the records that have DOIs that pasted in after manually creating the DOI using the Hakai Datacite client can not be updated.

This is problematic in that it if any changes to records that had DOIs manually created are made, those changes wont be automatically updated in Datacite.

We may need a function the check if a manually entered DOI exists in the Hakai Datacite client and if so then it should be updated. DOIs created outside of the Hakai Datacite client can not be updated automatically like this and would require a different manual workflow from the provider to update both at the same time, or have the two metadata records go out of sync.

[fostermh]

Is it not possible to update it on Datacite, or just not via the form? I ask because it could be that we are not storing the doi properly in the form if it's pasted in and that with more info, properly parsed, we could update from the form.

[Br-Johnson]

Its possible to update on Datacite no problem, just not from the Hakai metadata intake form.

I agree it probably just needs to be stored differently in the form if it's been pasted into the form manually

[timvdstap]

Upon some further reflection I might be leaning more towards auto-updating the DataCite record when saving changes made to a published record.

When a record is published in the Hakai Catalogue it'll already have gone through a round of review with Jessy/Brett/myself, and we'll also verify contents in DataCite when we change the status from Draft -> Findable. If people have the option of not updating the DataCite record when saving changes to a published record, don't we run the risk of having the metadata content in the Hakai Catalogue record be different from the metadata content in DataCite (and what if people simply 'forget' to sync the contents eventually?). Auto-updating the DataCite when saving changes in the form keeps the metadata record in the Hakai Catalogue as the single source of truth and keeps it simple?

Once a record is published in the catalogue and Findable in DataCite, we can't change it back to 'draft' in the metadata record anymore because this unpublishes the record and breaks the URL? So any changes would have to be made to a published record, where saving changes in the content also gets reflected in CKAN. I would imagine that we'd want to make sure those same changes are then also implemented in DataCite.

[JessyBarrette]

Here's a suggestion that could handle updates to published metadata records and DOI.

A metadata record can be in the following stages:

• Draft
• In review
• Published
• Modified

Once a record published, the DOI is also published.

Once this stage achieved, a metadata record can be modified via the form and saved locally. This record is then in the modified stage. Any changes to the form can only get reflected to the CKAN record and published when an update button is clicked within the form.

We can perhaps implement a check in the form that
warn the user if a form has been in modified for longer than a week.

Here's a diagram to summarize it all:
https://lucid.app/lucidchart/c3a7a532-8b0f-44bf-84ea-11c3630cda0d/edit?viewport_loc=-11%2C-28%2C1664%2C1965%2C0_0&invitationId=inv_c84a6d96-dcb3-41e0-aedf-f0bd6dac036b

[JessyBarrette]

Following our discussion, here's the latest workflow:

1. Draft DOI gets generated on click to Generate DOI
2. Draft gets updated on record submission
3. Published DOI gets updated on click to Publish within the form

[JessyBarrette]

#273 present the broader workflow which includes the CKAN record workflow at the same time.

[timvdstap]

edit - nevermind, steps already outlined in issue 273.

[JessyBarrette]

I believe this now fully implemented, my only question when is DataCite updated?

1. Only when the UPDATE DOI button is clicked
2. When UPDATE DOI is cliecked and save button?

[timvdstap]

Right now it's only when the UPDATE DOI button is clicked. I'm not sure whether it makes sense to have it automatically updated everytime you save your metadata record, but I think it would be great if the DOI is automatically updated upon submit.

[fostermh]

Right now it's only when the UPDATE DOI button is clicked. I'm not sure whether it makes sense to have it automatically updated everytime you save your metadata record, but I think it would be great if the DOI is automatically updated upon submit.

see Feature/update doi on submit