EarthCube FAIR Initiative

What is FAIR?

 

The FAIR Guiding Principles (Wilkinson et al, 2016) provide a frame for considering the state of human and machine readability of a dataset or resource, and the related metadata.

 

The FAIR Principles identify practices that reach across the data lifecycle, and include possible actions for researchers, data managers and stewards, as well as repositories of all types, to make data (or related) objects and resources *more* FAIR. That is, when assessing a data object or resource through this frame, the Principles should be considered on a continuum, that is a means toward full Findability – Accessibility – Interoperability – Reusability.

 

There are several key aspects that are core to FAIR operations:

• Findability requires unique, persistent identifiers, and these are registered or indexed to support search.

  • Data are described with rich metadata (Note that specific standards are not defined in the Principles; selection choices are left up to designated communities or domains to meet its particular needs.)

• Accessibility requires services that utilize Persistent Identifiers for retrieval, and that support authentication,  authorization, and controls for data usage license restrictions when needed. Service providers can also indicate the extent to which their underlying protocols are open and universally implementable.

  • The Principles specifically state that Metadata Records should persist, even if the data they describe are no longer available. (This provides continuity and persistence of the scientific record.)

• Interoperability requires the application of formal and accessible knowledge representation tools; those tools (such as standard vocabularies or ontologies) should also follow the FAIR principles

• Re-usability calls for the use of clear and accessible data usage licenses; and, that metadata include information on source and changes (provenance)

 

What FAIR is not: The FAIR Principles are silent on “open science” or “open data,” as well as on data quality. It should be noted that data that are compliant with the FAIR Principles and found to be FAIR can also be of poor quality, biased, or falsified. The FAIR community acknowledges that data may be “open as possible, as closed as necessary.” Data may be private, confidential, or proprietary, and under the FAIR Principles this is OK – especially when the aim is for metadata to be complete, fully available, and machine readable. Finally, successful application of the FAIR data principles is not, as yet, easily measured, though many tools are in development.