Re: Requirements for a possible "RDF 2.0"

On Jan 15, 2010, at 2:38 PM, Danny Ayers wrote:

> 2010/1/15 Pat Hayes <phayes@ihmc.us>:
>>
>> On Jan 15, 2010, at 4:57 AM, Danny Ayers wrote:
>>
>>> Jeremy, Toby, anyone,
>>>
>>>
>>> Aside from a little  tidiness, what would we actually gain through
>>> going the whole hog on what can go in which position in the triple?
>>
>> Well, the tidyness isnt to be sniffed at. It makes the specs a lot  
>> easier to
>> write (and to read), for a start. Also it makes syntax checking  
>> easier. It
>> makes it easier to think about the language (you don't have to keep  
>> mentally
>> checking yourself to see if you've broken some arbitrary rule.) IT  
>> makes the
>> closure inference rules and the query matching rules simpler and  
>> easier to
>> implement. And it does no harm.
>
> Ok, that's a good case.
>
>>> blank node predicate - what does that tell you that an rdfs:seeAlso
>>> wouldn't?
>>
>> SeeAlso tells you almost nothing, in fact.
>
> SeeAlso plus HTTP potentially tells you what you want to know.

Well, it might. But in fact it can be used for almost anything,  
according to the spec, and indeed it is.

>
> But you could for example have a
>> class of properties and say that one or more of them holds between  
>> two
>> things. Consider for example
>>
>> fatherOf rdf:type FamilyRelation .
>> motherOf rdf:type FamilyRelation .
>> sisterOf ... (etc.)
>>
>> and then
>>
>> _:x rdf:type FamilyRelation .
>> Alice _:x Bill .
>>
>> says that Alice and Bill are related somehow. (In case you are  
>> thinking that
>> subproperty would work here, it won't.)
>
> Would you actually want to model things like that?

People already do, eg expressing cladic relationships in biological  
taxonomies.

> The DL thing of
> individual vs. class goes underfoot for starters.

That is already abandoned in OWL 2, as it should have been long ago :-)

>
> hmm...it does seem a slight stretch - don't we have the rule of thumb
> that anything remotely significant deserves a name (URI) of its own?

Some people do, but Ive never bought it. And in any case, its a  
mistake to think of a bnode as 'thing with no name'. Its an  
existentially quantified variable, and those are very useful (and not  
at all mysterious.)

> Surely all predicates come under that umbrella.
>
> The make it neater/simplify argument works for me

thats my main one, yes.

Pat

> , not convinced by
> this example but I'll bear with you :)
>
>>>
>>> literal subject - aside from quotations:
>>>
>>> "I can't really see how it would be useful" <x:saidBy> <#me> .
>>
>> "37"^^xsd:number rdf:type PrimeNumber .
>>
>> "42"^^xsd:number :playsRoleIn  :HitchHIkersGuide .
>>
>> "66"^^xsd:number :greaterThan _:x .
>> _:x :age :PatHayes .
>>
>> (which is why I don't get a full SS pension this year).
>
> Well I never had you down as a Schutzstaffel kinda guy...

After all the work Ive done for them, they owe me :-)

>
>>
>> I'm sure I could think of others.
>>
>> But the main point is that allowing all this makes the language  
>> *simpler*.
>> the question should be, why the hell did we not allow it in the  
>> first place?
>
> Makes a lot of sense to me, I like tidy, thanks Pat.
>
> Cheers,
> Danny.
>
>

------------------------------------------------------------
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

Received on Friday, 15 January 2010 21:38:29 UTC