FAIR Guiding Principle F1:
(meta)data are assigned a globally unique and persistent identifier
Interpretation of F1
Principle F1 states that digital resources, i.e., data and metadata, must be assigned a globally unique and persistent identifier which serves as a permanent machine interpretable reference. The GO FAIR Foundation emphasises the need for persistence and global uniqueness, as well the property of resolvability of the identifiers (see also A1). Globally unique means that the identifier is guaranteed to unambiguously refer to the intended resources (where 'global' is intended to mean 'universal' as there are described digital assets outside the 'world'). Therefore, it is insufficient for it to be unique only locally (e.g. unique within a single, local database). Persistence refers to the requirement that this globally unique identifier is never reused in another context, and continues to identify the same resource over time, even if that resource should no longer exist, or moves from one digital environment to another. While global uniqueness is a technical property (i.e., an algorithm that can guarantee with mathematical precision that the issued identifiers are unique), persistence is a social commitment made by the stakeholder responsible for issuing the identifiers, that these identifiers will continue to map to the objects they identify for a defined period of time. An additional property supported by the GO FAIR Foundation is that the identifier is also ‘resolvable’ by machines. An identifier is most useful in a large-scale automated environment only when it can be resolved into (i.e., linked to) the object it identifies. Furthermore, the GO FAIR Foundation also assumes predictable identifier resolution behavior, allowing identifier resolution to behave consistently across multiple requests. Taken together, the GO FAIR Foundation assumes FAIR implementations to have Globally Unique, Persistent and Resolvable Identifiers (GUPRIs).
This interpretation of F1 is based on 'FAIR Principles: Interpretations and Implementation Considerations'. Jacobsen et al, Data Intelligence 2020; 2 (1-2): 10–29. doi: https://doi.org/10.1162/dint_r_00024