Metadata Element for Long-term Preservation

[kpletsch]

What is the problem that your suggestion solves?

A lot of research data is currently not preserved (dpc bitlist, [http://doi.org/10.7207/dpcbitlist-23, see published and semi-published research data]) even if the data is published, citable and part of research communication and, accordingly, should be accessible and reusable in the long-term. There is no central database where funders, researchers or institutions can query which research data publications have been archived and preserved by which responsible party. Still, funders increasingly ask for publication of research output in “trusted data repositories” indicating the importance of long-term accessibility and reusability, beyond life time of personnel, projects, software, format standards and hardware. On the other hand, Crossref, Keepers Registry and library catalogues allow such documentation but these three solutions don’t cover research data and, as such, don’t provide solutions in this case.
Such a metadata element would

1. allow institutions to promote which research data publications they have archived,
2. allow funders to check if funding obligations have been fulfilled,
3. would allow the research community and research infrastructure community to share responsibility for the considerable task of preserving research data in an informed manner and
4. may allow researchers to check which data has been archived and preserved already.
   Information about the validity of certification ensures that the respective institution actually still exists and still preserves and curates data they are responsible for in a responsible manner. It ensures that preservation solutions are actually trusted repositories as recommended by funders.

What solution might meet your needs?

Addition of an optional metadata element isArchivedBy for research data publications, i. e. connected to a research data publication DOI. The value should be a machine readable identifier of a data repository, a data provider, a service provider or institution which has an active digital long-term preservation certification (CoreTrustSeal, nestor seal or ISO 16363) and therefore conforms to the long-term preservation values specified by ISO 14721.
Specifically the value should be/have:

• machine readable identifier of a certified data provider, of a certified service provider, of a certified data and service provider, of a certified institution. (long-term preservation focused certification like CoreTrustSeal, nestor seal)
• AND this machine readable identifier should be provided by a registry that also contains certification information linked to the data provider, service provider, institution.
• AND the record of the data/service provider or institution with its machine readable identifier and its certification information can be queried by registry API, which may allow automated updates to the DataCite record in the future.
• AND the registry should allow documentation of various kinds of certification, which would support future development of this functionality.
  An example for such a registry is re3data - the re3data record already contains information about certification of the institution and it also has an API [https://www.re3data.org/api/doc]. An API request may use the re3data ID of e. g. World Data Center for Solid Earth Physics (r3d100010447) in a request URL [https://www.re3data.org/api/v1/repository/r3d100010447] and the response contains details about the service, including its certification.
  The specific metadata element can have a slightly different label, e. g. isArchivedAt, isPreservedBy or similar. Schema.org supports the following markup properties: archivedAt [[https://schema.org/archivedAt],(https://schema.org/archivedAt)], conditionsOfAccess [https://schema.org/conditionsOfAccess], holdingArchive [https://schema.org/holdingArchive]. Of these only archivedAt and conditionsOfAccess are part of the dataset markup of schema.org [https://schema.org/Dataset]. These properties have different intentions compared to the one proposed for DataCite, i. e. allowing documentation of links to archived versions or documentation of access methods, both in unstructured free text. They don’t allow querying responsibility in a structured manner and don’t cover dark archives. Dark archives don’t allow access to archived versions and work in concert with access platforms, meaning they don’t provide an URL to the archived version.
  Additionally – maybe as a future implementation – the addition of the type of certification of the institution as a sub-element in DataCite would be beneficial. And finally, the addition of another sub-element covering the validity of the certification would be valuable – since re3data does not contain this information at the moment, this may only be possible in the future.

Your name

Katja Pletsch

Your organization

ZB MED – Information Centre for Life Sciences

What alternatives have you tried or considered?

There is currently no workaround. In case a German regional library catalogue supports research data cataloguing (few officially and structurally do) and in case an institution catalogues their research data and in case the institution adds preservation information to the library catalogue some research data preservation status may be queried via these. This is insufficient due to implementation level and limited reach/coverage since research is international.
Another workaround would be looking up each institution with a certification on re3data and on coretrustseal.org individually, analysing their web presence for APIs like OAI-PMH and analysing their content for research data. As this involves several steps, might be dependent on institutional infrastructure and does not allow structured querying across institutions this workaround does not serve as a realistic alternative.

Is there anything else you would like to share?

No response

What group(s) would benefit from your suggestion?

• DataCite members registering DOIs and metadata and administering accounts
• Developers building integrations with DataCite services
• Harvesters who index DataCite DOIs and metadata
• Individuals or institutions analyzing DataCite DOIs and metadata
• DataCite staff
• Other

If other group(s), please describe.

Researchers, Funders

[KellyStathis]

@kpletsch Thank you for this suggestion, and I'm sorry for the delay responding!

Based on your description, I understand it's important to be able to link to the institution or repository that archives the data, although there won't necessarily be a direct link to the archived copy of the data itself (e.g., in a dark archive).

Therefore, I would suggest using Contributor with contributorType "HostingInstitution" for this use case. This fits with the definition, which has a note that it can be used for an organization storing data offline:

HostingInstitution
Description: Typically, the organisation allowing the resource to be available on the internet through the provision of its hardware/software/operating support.

Usage Notes: This role normally falls on the University, research center or organization where the data center/data repository belongs.

Example: Université Grenoble Alpes (UGA)

May also be used for an organisation that stores the data offline - often a data centre if that data centre is not the “publisher” of the resource.

Would this potentially work for your use case?

[Ka-Markus]

@KellyStathis, I co-wrote the suggestion so I’ll answer in @kpletsch’s stead. Sorry for the delayed response as well!
We took hostingInstitution into account but discarded it for our purpose since it isn’t specific enough.
It does not allow separation of storage/hosting and digital preservation and we need the ability to filter specifically for preserved data. Digital preservation is not storage as it includes active preservation of data beyond changes in technology, standards and formats. It depends on clearly defined and transparent long-term responsibility following standards like ISO 14721, DIN 31644, CoreTrustSeal and nestor seal certification. Consequently and in order to clearly separate this responsibility from simple storage we wanted to reference a certified repository or archive in digital preservation queries. Crossref also has a distinct metadata element for preservation and even monitors trusted archive providers: https://www.crossref.org/documentation/schema-library/markup-guide-metadata-segments/archive-locations/.
Another specific scenario that this new element would cover is the Dark Archive scenario (content in a Dark Archive is not accessible via internet). This is in particular helpful for those cases, where an outward facing repository, which the DOI points to, is provided by a different entity than the digital archive. The DataCite record could then be used to reach out to an alternative source in cases where the original repository is no longer available.
Additonally, hostingInstitution seems to generally focus on availability via internet, which is not necessarily a focus in digital preservation.
Furthermore, we did not want to refer to an institution since an institution may offer a repository without digital preservation but also a digital archive with preservation (or another setup where only some of the hosted material is preserved). Not all data at an institution necessarily is transferred into a preservation service/digital archive and it needs to be clear which is.

Therefore we wanted to suggest a metadata element that is distinct from documenting who stores or provides internet access to data, that includes certification information to make sure digital preservation is actually provided and that allows documentation of digital preservation by specific services and not institutions.

[aquast]

1+