Re: test case A and superProperty

>As per my recent posts (sorry for the shotgun effect, I wanted to 
>keep the threads separate), its been suggested that we can have test 
>case A be yes and test case C be no.
>
>I just wanted to note that there is an interaction between this test 
>case and rdfs:subPropertyOf.  Consider:
>
>   _:a dc:title "4th July" .
>   _:b dc:date  "4th July" .
>
>[I don't know DC well enough to know whether this is a realistic 
>example, I suspect not so consider this just illustrative]
>
>All is reasonably well.  With appropriate datatype range constraints 
>one of these denotes a string and the other a date.
>
>Now add a common super property for all the dc properties:
>
>   dc:property rdf:type        rdf:Property .
>   dc:title rdfs:subPropertyOf dc:property .
>   dc:date  rdfs:subPropertyOf dc:property .
>
>This now entails:
>
>   _:a dc:property _:l .
>   _:b dc:property _:l .
>
>I assume this is still monotonic, since I haven't actually retracted 
>anything, though it sure feels like changing my mind to me.

My first reaction was, what's the problem?? Technically this is all 
consistent. But thinking about it more, I see that there is something 
odd about this. One would reason as follows: the *reason* why 
dc:property holds for both literals is different in the two cases, so 
even though its the same property, the literals aren't being treated 
the same (assuming that the literals 'mean' the same thing in the 
superproperty triples as they did in the corresponding subproperty 
triples), so the common-b-node inference is no more justified in the 
superproperty case than it was in the subproperty. Hmmm, yes, good 
point.

>A more extreme form is to postulate the existence of a common super 
>property of all properties, which we have talked about in the past. 
>If there were such a thing and the answer to A is yes, then we'd 
>have tidy literals anyway.
>
>At this point I think I need to hear Pat say, "relax".

Well, after the F2F discussion and subsequently, I think that this 
compromise idea (with literals being syntactically tidy but 
semantically untidy) isn't really worth the trouble it causes. If 
literals are going to be semantically untidy then lets have them be 
syntactically untidy as well, since that keeps everything coherent. I 
think we should stop trying to be clever, and just decide one way or 
the other: either literals just are string-ish things which denote 
themselves, or they are genuine names which can denote anything, 
depending on the datatyping applied to them; and in the latter case, 
then really should not be syntactically tidy, since different 
datatyping contexts have to be allowed to attach to distinct literal 
nodes. I myself would prefer the latter, and also to allow literals 
to be in the subject position, but I gather that this isn't 
acceptable; so I would vote that we stick to the other simple option 
and reject implication A.

Pat

-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Wednesday, 10 July 2002 14:03:47 UTC