Re: Well Behaved RDF - Taming Blank Nodes, etc.

On 12/19/2012 7:53 AM, Atila wrote:
> On 18-12-2012 15:11, Lee Feigenbaum wrote:
>> For the latter, I'd probably just model it as :Jane :wants :Apple or 
>> something like that.
> Hum, and if she wants a green apple, like
>
> :Jane :wants [a :Apple; :color :Green]
>
> ?

Well, if my intuition is right that

:Jane :wants [a :Apple]

is incorrect modeling for "Janes (notionally) wants an apple", then the 
above is similarly incorrect for Jane (notionally) wanting a green 
apple, and instead I'd suggest:

:Jane :wants :GreenApple .
:GreenApple rdfs:subClassOf :Apple ; rdfs:subClassOf [ a owl:Restriction 
; owl:onProperty :color ; owl:hasValue :Green ] .

(oh look, a blank node! :-) )

> I think that the botton line is that machines doesn't need blank nodes 
> at all, they can work very well without it. But humans do.
> It is possible to express things using URIs to everthing, but it is 
> just too messy in some cases. It would be painful, specially to 
> beginners.

Blank nodes are subtle and non-obvious (see the above discussion). I 
disagree that they're simpler for beginners. I think beginners would 
have an easier way getting started without them.

> When i first heard about RDF, I got scared that every single thing 
> would got to have an URI. I thought "that can describe social networks 
> well, but will be useless to NLP", then I heard about blank nodes and 
> saw I was wrong.
>

We do NLP all the time and everything gets a URI.

Lee