On Nov 16, 2012, at 10:34 AM, John F Sowa wrote: (01)
As I said, I am ignoring all discussions of history.
... (02)
> JFS
>>> An even better notation would used typed triples, such as
>>>
>>> {T1:V1, T2:V2, T3:V3}
>>>
>>> This could be translated to RDF, but it would expand to five triples.
>>> So the type labels often get lost along the way.
>
> PH
>> What?? The Web does not use a "type" to specify what a URI points to.
>> URIs are Web addresses in themselves (sometimes a bit more, but
>> *at least* a Web address.)
>
> As Duane N and Doug F have pointed out, the original URLs are primarily
> addresses with a bit more information given by the "http" prefix. Since
> the original WWW assumed that the browser's primary goal was to display
> whatever occurred at the target of a URL, any descriptors in a document
> would tell the browser how to display the content.
>
> But a reasoning system needs more information about how it should
> treat a name.
>
> PH
>> What 'types' are you talking about here?
>
> Two kinds of information:
>
> 1. Syntactic information that says some occurrence of a name should
> be treated as a reference to a function, relation, or individual. (03)
RDF has this. The RDF semantics follows the CL model theory in allowing any
name to refer to an individual and also to a unary or binary relation (called
there a class and property, respectively). The RDF triple syntax and the use of
rdf:type provides the syntactic machinery to determine which of the three uses
is intended by each occurrence of a URI in RDF. (04)
RDF does not currently allow relations of higher arity (I am trying to get this
fixed, but don't hold your breath) and it treats functions as functional
relations (and OWL provides a vocabulary for describing these.) (05)
> In RDF, it's not clear whether a name refers to a document, to
> its content, or to an operation specified by its content. (06)
RDF is quite clear on this: RDF semantics completely ignores what the URI
resolves to using HTTP (or indeed any other xxTP). This was an early, and quite
conscious, design decision. So we have what a URI denotes, and what it
"identifies" (in the current W3 jargon) and these are (ab initio) independent
of one another. Users may choose to define whatever relationship they feel
most useful, varying from nothing to identity. Back in 2003, it was felt (IMO
correctly) that to legislate on this would be counterproductive until a body of
usage experience had been accumulated. Sure enough, several divergent views are
emerging. (07)
The notorious http-range-14 decision made this depend upon the HTTP return code
(200 or 303: basically, if you want to talk about a Web "document", use the
identifying URI to refer to it; otherwise, the source should generate a 303
redirect), which I howled against as insanity when it was proposed, but which I
confess has beeen found to work quite well in practice, cf. DBpedia. The
current "linked data" view is that the URI should resolve to a source of
information about the thing denoted, typically more data. (08)
>
> 2. A monadic relation that specifies the ontological category or
> position in a type hierarchy. In CL, that information can be
> stated in relational form (T1 V1) or in a quantifier restriction. (09)
And in RDF it is expressed using rdf:type, provided for exactly that purpose.
RDFS (or OWL if you are more ambitious) lets you describe the actual hierarchy. (010)
Pat (011)
>
> John
> ______________________________________________________________________
>
> References
>
> 1. The original DAML proposal from February 2000:
> http://www.w3.org/2000/01/sw/DevelopmentProposal
>
> 2. The list of 22 DAML groups and their members:
> http://www.daml.org/researchers.html
>
> 3. (a) The W3C version of the DAML final report:
> http://www.w3.org/2005/12/31-daml-final.html
>
> (b) The official final report as delivered to DARPA:
> http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA458366
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/
> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>
> (012)
------------------------------------------------------------
IHMC (850)434 8903 or (650)494 3973
40 South Alcaniz St. (850)202 4416 office
Pensacola (850)202 4440 fax
FL 32502 (850)291 0667 mobile
phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes (013)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (014)
|