"characteristic of" (RO:0000052) textual and logical definition inconsistencies that are related to COB intended meaning

[StroemPhi]

initially asked on Slack by me :

Hi, is there a reason why 'characteristic of' does not have SDC as a domain restriction? The textual definition is:

a relation between a specifically dependent continuant (the characteristic) and any other entity (the bearer), in which the characteristic depends on the bearer for its existence.

And in the inverse relation 'has caracteristic' the range is restricted to SDC.

@cmungall replied

the definition was written pre-COB. The definition should be rewritten to “a relation between a characteristic and any other entity (the bearer)“. But characteristic is indeed declared equiv to SDC in cob-to-external. This whole area is a terminological mess. The root note of PATO is named “quality” but that decision was made before BFO made restrictions on quality, PATO:quality is really characteristic, as cob-to-external makes clear

Then I looked more closely into the issue and made these observations:

1. If I understand @cmungall's comment (👆) correctly, the characteristic in the current textual definition of RO:'characteristic of' should actually, within the COB context, be understood as the original PATO:quality, which is not BFO aligned and thus no BFO:SDC, and, which would render the current textual definition of RO:'characteristic of' incorrect. Hence, your rewrite proposal.
2. Currently however, PATO_in_RO:quality (via OLS) is subsumed under BFO:SDC, which causes the following to hold true:
   • This use of PATO_in_RO:quality is in line with the textual definition of RO:'characteristic of', as well as with the logical definition of its inverse RO:'has characteristic' and, by the means of backlinking to RO:'characteristic of', also with the inverse's textual definition.
   • So, to remain consitent within the current RO, proposing to assert BFO:SDC also as the domain of RO:'characteristic of' does make sense.
     3. Now, if you were to fully implement the intended/planned meaning from 1. in COB, that is to switch to what you seem to have discussed in #284 and in COB#65, the following edits would have to be made, given my reasoning is sound and knowledge complete enough:
   • In COB, assuming a build-from-tsv pipeline, cob-to-external.tsv#L121 (COB:0000502 owl:equivalentClass BFO:0000020) would have to be removed.
   • And in RO, as COB currently imports the RO:'has / characteristic / of' relation pair, you would have to:
     • change the textual definiton of RO:'characteristic of' to the proposed "a relation between a characteristic and any other entity (the bearer)" ,
       • OR, to stay in line with the current definition of PATO:quality, sharpen this new definition to "a relation between a characteristic (a dependent entity) and any other entity (the bearer)"
     • remove the SDC range constraint on RO:'has characteristic', to remain coherent
     • think about ontologically committing by adding PATO:quality as domain respectively range constraing to this relation pair
     • move PATO_in_RO:quality somewhere else in the BFO hierarchy than under BFO:SDC
       • But where should that be? If dependent entity is to be understood in a BFO context as the union of BFO:DC and BFO:SDC,
       • BFO1.1:'dependent continuant' would have to be revived to serve as parent. 😟
       • BFO1.1:'dependent continuant' could then be asserted as domain respectivly range contraint on the RO:'has / characteristic / of' relation pair, if you wanted to ontologically commit to this.

[StroemPhi]

And RO_in_COB:'characteristic of' is not declared functional although it should be according to the docs and RO:'characteristic of'.

[matentzn]

@StroemPhi you raise a lot of great points..

I just wanted to wish you strength and patience, wholeheartedly, and recommend that you break this ticket down into as small pieces as possible. Read issues like #452 or #341 for your own pleasure.

Maybe we could:

• rename PATO:quality to characteristic
• Leave this class uncommitted wrt to BFO
• and have it be the domain of characteristic of

I think the biggest problem with COB is (what makes it so slow) is that we want to align it with BFO. Balancing the messiness of biology (disease as material entity vs SDC vs process vs disposition) with the idealistic vision of assigning everything to BFO for some artificial ontological clarity needs to shift towards the former.

[StroemPhi]

@matentzn I tried to break all that I could grasp down to individual pieces with the formating. What you propose is essentially what I wrote in 3. apart from the alignemt with BFO. And I guess "2.", the missing domain contraint on RO:characteristic of', is not so important at the moment, as it is implied via the inverse relation anyway.