Re: Contradicting definitions of “property”

On Aug 23, 2012, at 5:25 AM, Antoine Zimmermann wrote:

> Le 23/08/2012 07:55, Pat Hayes a écrit :
>> On Aug 22, 2012, at 4:58 PM, Richard Cyganiak wrote:
>>> On 13 Aug 2012, at 18:29, Antoine Zimmermann wrote:
>>>>> Note that the new Introduction section in the RDF Concepts ED
>>>>> contains an *informative* sentence that introduces the term
>>>>> “property” [4], and it is in line with RDF Semantics:
>>>>> [[ The predicate itself is an IRI and denotes a binary
>>>>> relation, also known as a property. ]]
>>>> This is not what the RDF semantics says. A predicate denotes a
>>>> resource that must be in IP, the set of properties in the
>>>> interpretation. Resources in IP are associated with a binary
>>>> relation via the extension function IEXT. This is an important
>>>> distinction since this is what allows RDF to talk about
>>>> properties, classes, etc as instances.
>>> Ah, right. I forgot about the class/property extension stuff in RDF
>>> Semantics.
>>>> If predicates were denoting binary relations, the following would
>>>> be RDFS-inconsistent, when it is, in fact, RDFS-consistent:
>>>> :p  rdf:type  xsd:string . :s  :p  :o .
>>> Do I get this right? This would be inconsistent because the first
>>> triple says its a Unicode string, and the second triple entails
>>> that it is a property, and hence (if my phrasing above were indeed
>>> correct) a binary relation. And a Unicode string is not a binary
>>> relation.
>>> And in reality, as RDF Semantics defines things, the second triple
>>> only entails that the Unicode string *has a property extension*,
>>> and the property extension is a binary relation. Hence, no
>>> contradiction. Anything can have a property extension.
>>> Right?
>> Right. In RDF, in fact, everything *does* have a property extension
>> (whether you are using it or not, it is there to be used.) In this it
>> follows ISO Common Logic, by the way.
> Nope. From RDF Semantics 2004, Section 1.3:
> "3. A mapping IEXT from IP into the powerset of IR x IR i.e. the set of sets of pairs <x,y> with x and y in IR ."
> only elements of IP, aka properties, have a property extension.

Ah yes. Whoops. I have been talking to CL folk about CL so much lately that I forgot that we wimped out on this point in the RDF semantics :-)


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

Received on Thursday, 23 August 2012 15:03:16 UTC