Represent the metaphysics, not the categorisation logic itself

[pbuttigieg]

Branching off from #1578 (comment)

This is a discussion thread to log a recommended practice of always - at least in realist ontologies - representing the metaphysical entities rather than the quirks and limits of a categorisation or classification system.

For example, the IUCN case of Ecosystem Functional Groups (EFGs; discussed in #1578) presents a concept - the EFGs - which are groups of ecosystems with certain traits and attributes. The ecosystems certainly exist, their functions (in a BFO sense) also exist, but the grouping of ecosystems is more by human fiat than through any metaphysical connectedness.

A realist ontology should avoid creating classes or instances of fiat-driven entities, unless they are represented as such via fiat boundaries and parthood.