This is work in progress and needs discussion.
Here is a visual representation of this pattern, using the following notations and conventions:
The figure shows a context, in which an agent of a party that perceives the context (and its contents) uses identifiers for the purpose of identifying entities within that context. Typical contexts are those in which a computer system/application acts as such an agent, but any other situation where (an agent of) a party is acting has its own particular context. The context is significant, because within that context, not every entity that a party knows to exist will be (expected to be) present, or needed to be identified.
It is assumed that identifiers are associated with specific classes of entities. Such associations may differ between contexts. For example, in the context of discussing war planes, the identifier
B52 refers to a class of bomber planes, whereas in the context of Schiphol Amsterdam airport,
B52 would refer to a (boarding) gate, and in the context of drinks, it would refer to a specific cocktail. Examples in the IT include identifiers in the syntax of e-mail addresses, which may be used as an email address, as a username for logging in, as a SIP address (for making calls over the Internet), etc.
Typically, identifications for which identifiers are used are limited to specific contexts where the actor that executes the identification-act knows the kind of entity that needs to be identified, and when using some identifier, it would expect it to be useable for instances of such a kind.
The figure introduces the concept
ID, the idea of which is that it is a composition of whatever it takes in terms of identifiers to be able to identify an entity of a specific type in a specific context. In the simplest case (in a small context, without many entities to choose from), a single identifier may suffice - specifying the entity type may not even be necessary. In other cases, a single identifier m ay not suffice, and multiple identifiers are necessary
Text to be added from here on. Thoughts revolve around changing 'entity type' to 'attribute type' or such. That might result in a mapping (Entity, Attribute Type, Identifier) in the current figure to (subject, predicate, object) or pehaps better: (subject, key, value), i.e. (key, value) pairs for an implied subject. That could form the basis of doing identification in the data realm (which is currently still lacking)
Earlier ideas for this can be found in "Putting Identifiers in the Context of eHealth"