First published in 2016, the FAIR Principles were developed as a concise and measurable set of guidelines to ensure that research (meta)data are findable, accessible, interoperable, and re-usable by both humans and machines, thereby maximizing their scientific value (Wilkinson et al., 2016). Since their introduction, the FAIR Principles have become a standard by which research data management tools and services can be evaluated and have seen rapid adoption by funders, publishers, and service providers (Wilkinson et al., 2018).
FRDR supports the FAIR Principles in the following ways:
F1. (meta)data are assigned a globally unique and eternally persistent identifier.
FRDR assigns DOIs to all accepted submissions through DataCite Canada.
F2. data are described with rich metadata.
All content deposited directly into FRDR is required to conform to a template from which complete Dublin Core and DataCite XML can be generated; custom metadata fields are supported when applicable with the goal of ensuring these map to existing standards.
F3. (meta)data are registered or indexed in a searchable resource.
FRDR's own OAI endpoint is used to harvest metadata for discovery in FRDR's index of Canadian data repositories, and this OAI endpoint is open to the public.
F4. metadata specify the data identifier.
DOIs and other identifiers are visible on objects' landing pages.
A1. (meta)data are retrievable by their identifier using a standardized communications protocol.
DOIs are indexed by aggregators such as re3data, and DOIs can be used as an entry point to FRDR's own index.
A1.1 the protocol is open, free, and universally implementable.
DOIs and OAI are open standards. Additionally, the code for FRDR's harvester is openly available: https://github.com/axfelix/globus_oai
A1.2 the protocol allows for an authentication and authorization procedure, where necessary.
Globus authorization is implemented for any datasets which are not publicly available, and ORCiD can be used to register for authorization.
A2 metadata are accessible, even when the data are no longer available.
FRDR has been designed to maintain tombstone pages for datasets which have been removed from FRDR itself.
I1. (meta)data use a formal, accessible, shared, and broadly applicable language for knowledge representation.
FRDR has been working with Portage's Discovery Expert Group to facilitate the use of OpenRefine to map FAST vocabulary terms onto harvested keyword metadata.
I2. (meta)data use vocabularies that follow FAIR principles.
The FAST vocabulary is consistent with FAIR principles.
I3. (meta)data include qualified references to other (meta)data.
Wherever possible, FRDR's curators make use of "related" elements in submitted datasets in order to improve data linking.
R1. meta(data) have a plurality of accurate and relevant attributes.
FRDR's default metadata template based on the Dublin Core and DataCite schemas is robust, with many accurate and relevant attributes from which users may choose.
R1.1. (meta)data are released with a clear and accessible data usage license.
FRDR defaults to, and encourages the use of, Creative Commons licenses wherever possible for submitted data.
R1.2. (meta)data are associated with their provenance.
Provenance is automatically recorded from curator logs during our approval process and permanently associated with deposited data.
R1.3. (meta)data meet domain-relevant community standards.
In addition to implementing Dublin Core, DataCite, OAI, and DOIs as above, repositories are approached when issues with metadata arise so corrective action can be taken.
See also: https://www.go-fair.org/fair-principles/
Wilkinson, M. D., Dumontier, M., Aalbersberg, Ij. J., Appleton, G., Axton, M., Baak, A., … Mons, B. (2016). The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data, 3, 160018. https://doi.org/10.1038/sdata.2016.18
Wilkinson, M. D., Schultes, E., Olavo Bonino, L., Sansone, S-A., Doorn, P., & Dumontier, M. (2018). FAIRMetrics/Metrics: FAIR Metrics, Evaluation results, and initial release of automated evaluator code. Zenodo. https://doi.org/10.5281/zenodo.1305060