RO axioms that have domain/range constraints with BFO classes above the COB shoreline

[cmungall]

What do we do here?

1. Leave as dangling
2. rewrite using UnionOf
3. merge in the BFO abstract parents
4. move towards specific relations in RO

1 is highly unsatisfactory, the way RO is structured we would lose a lot of massively useful QC

2 is formally an overreach, since BFO doesn't (AFAIK) include closure axioms justifying this. It also means that we lose QC for EL++ (though I think Elk can handle UnionOf?)

3. Is the most practical and likely a good first pass. It does pose the question of what it means to use COB. At least in the first pass using COB will mean nothing more than only using centrally sanctioned injections (Define if and when it is OK to inject axioms into an external ontology OBOFoundry.github.io#1443) which is a good thing. This is quite conservative. But perhaps fine.

To properly implement 3 we may need to make a BFO subset that is the above the shoreline cut

[wdduncan]

@cmungall For (3), are proposing that BFO classes such as continuant and occurrent be included in COB?

[cmungall]

@wdduncan

For 3, the proposal would not be to formally include them in COB, but we would have a release artefact that includes BFO classes above the shoreline like continuant and occurrent (i.e. superclasses of existing COB-BFO equivalents, which corresponds to domains and ranges in ro-core)