W3C home > Mailing lists > Public > www-rdf-logic@w3.org > November 2001

RE: literals and typing

From: Pat Hayes <phayes@ai.uwf.edu>
Date: Tue, 13 Nov 2001 21:01:22 -0600
Message-Id: <p0510104bb8178e19af7f@[65.212.118.147]>
To: "Geoff Chappell" <geoff@sover.net>
Cc: www-rdf-logic@w3.org
>  > -----Original Message-----
>>  From: www-rdf-logic-request@w3.org
>>  [mailto:www-rdf-logic-request@w3.org]On Behalf Of Pat Hayes
>>  Sent: Friday, November 09, 2001 4:18 PM
>>  To: Geoff Chappell
>>  Cc: www-rdf-logic@w3.org
>>  Subject: Re: literals and typing
>>  >>
>>  >>  To elaborate on the above, BTW, I just posted a longer comparison:
>>  >>  http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0295.html
>>  >>
>>  >
>
>[...]
>
>>
>>  Well, yes; but all the proposals have snags, unfortunately. I wish
>>  there was a clear winner, but I think we will have to compromise
>>  somewhere. I personally would love to see something like X, in its
>>  stark simplicity, but I doubt if the RDF core WG will be willing to
>>  adopt a new, unproven, URI scheme as part of the standard. And the P
>>  schemes are rather 'delicate' in that they require datayping schemes
>>  to be carefully crafted to be upward compatible, and will completely
>>  fail if they are not.
>
>If I understand the delicacy issue with P/P++ it's that a class and one of
>its subclasses might have different lexical domains (e.g. hexint, int) and
>so it will be unclear/ambiguous in which form the literal value is actually
>encoded as a string. But doesn't the same issue exist with S? if we have:
>(#whoknows hexint "70") and (hexint subPropertyOf int) we can infer
>(#whoknows int "70"). Can't these problems exist in any of the schemes
>except X (and only not there because types aren't exposed to inference -
>i.e. there are no datatypes visible to RDF).

The issue is not so much the possibility of stating daft things, but 
the impossibility of preventing an RDFS reasoner from drawing bad 
conclusions. Its true that the S scheme allows one to state 
incompatible datatyping, but if you do not do that, then things will 
work. But if the datatype class heirarchy has incompatible lexical 
mappings on a supertype, then even a correct assertion of datatyped 
range information can produce an incorrect answer, using valid RDFS 
class reasoning, since all superclasses of a range are also ranges.

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 Tuesday, 13 November 2001 22:01:21 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 5 February 2014 07:16:02 UTC