Data Repository Evaluation - Are the FAIR Data Principles fair?

[ghost]

Since the work of 4TU is mentioned under 'Measuring Change' on the about page of this knowledge pool, I would like to quickly link to the corresponding work.

• Pre-Print Practice Research Paper "Are the FAIR Data Principles fair?"

• Excel Spreadsheet with the evaluation, statistic and graphs of 37 dutch research data repositories / archives (also including Figshare, Mendeley Data, EUDAT B2Share, Zenodo)

• Blog-post connecting all the dots for the IDCC 2017

[sjDCC]

Perfect, thanks @jkb4TU

[CaroleGoble]

The EU Summit's [FAIR Certification] Intent Raises concerns with me.
Stewardship should be within the domain using accepted community standards.

A librarian viewpoint of FAIR data is quite different to a Biologist reusing a dataset. A colleague who runs a major data archive widely used was accused of not have good metadata to put into OpenAIRE. In fact their dataset was highly curated and excellent and compared to the majority of data in OpenAIRE was exemplary. But one field considered important for EU accounting was missing. Which was the most FAIR?

it reminds me of Dickens Hard Times: " An exchange between Gradgrind, Bitzer and Sissy Jupe over the proper definition of a horse. Bitzer, who has learned a definition by rote, classifies it as a "Quadruped" and "Gramnivorous," whereas Sissy, the horse-breaker's daughter is reprimanded for possessing "no facts, in reference to one of the commonest of animals" (https://omf.ucsc.edu/london-1865/schools-and-education/victorian-education.html)" though clearly Sissy knows infinitely more about horses than Gradgrind who merely knows the label.

[sjDCC]

Is this the text you're referring to @CaroleGoble

[Accreditation/certification] Scientists must be assured that the European and national scientific research infrastructures where they deposit/ access data conform to clear rules and criteria (e.g. certified) and that their data is FAIR compliant. An accreditation or certification mechanism must be set in place based on agreed processes and an accreditation or certification body must maintain an up-to-date and accessible catalogue of certified repositories. Experience from existing accreditation processes must be taken into account.

[sjDCC]

I've just been reviewing the FAIR metrics site and from the diagram it looks like they're considering metrics on a domain basis. I agree that this is key because what constitutes good metadata comes down to the specific of each community.

I know there's a lot of desire for recommending certified repositories, but I think we need to exercise some caution here and be pragmatic about the shift that's required. Very few repositories have certifications currently and it's not a non-trivial task to obtain them. There are some excellent repositories that aren't certified too. The 4TU study pointed out a few simple steps that can make repository much more FAIR and I think that's the kind of low benchmark we want to start with if we're requiring it of all providers. Where I do agree with this statement is that we need to build on existing accreditation processes, not start something completely new.