Re: bang ! in turtle

On 01/03/11 17:39, Nathan wrote:
> Andy Seaborne wrote:
>> On 01/03/11 17:23, Nathan wrote:
>>> Andy Seaborne wrote:
>>>> On 01/03/11 16:41, Nathan wrote:
>>>>> actually ^ may be better.. such that
>>>>>
>>>>> :a :b :c .
>>>>>
>>>>> could be written as:
>>>>>
>>>>> :c ^:b :a .
>>>>>
>>>>> meaning
>>>>>
>>>>> :c [ owl:inverseOf :b ] :a .
>>>>
>>>> meaning there is a there is bnode in the predicate position.
> ...
>>>>
>>>> SPARQL has:
>>>>
>>>> :c ^:b :a .
>>>>
>>>> meaning
>>>>
>>>> :a :b :c
>>>>
>>>> reverses subject and object. The matching process really does swap
>>>> subject and object.
>>>>
>>>> http://www.w3.org/TR/sparql11-query/#propertypaths
>>>
>>> wouldn't that require subjects as literals?
>>
>> Yes.
>
> hmm, this is probably a fruitless conversation then, because we can't
> change the RDF semantics, and the choices are bnode predicates or
> literal subjects to allow this (and to align w/ sparql I guess..).

Why?  Treat it as a syntax issue.

Just ban triples with literal-as-subjects as an error during the step 
that turns syntax into a set of triples.  After all, it's the syntax in 
RDF/XML that is the main block on literals-as-subjects anyway.

:a ^:p :b . is legal.

:a ^:p "foo" . is not legal.

It might even be possible to enforce in the grammar itself although that 
will depend on what other shorthand notations get addded.

(wandering further afield ...)
At the moment there is shorthand for common subject.

What about common object:

:a :b :c .
:x :y :c .
==>
:c ^:b :a ;
    ^:y :x .

and common subj/obj:

:a :b :c .
:c :p :x .
==>
:c ^:b :a ;
    :p :x .

but that could be:

:a :b :c :p :x .

but whether that's a *good* idea is quite another matter.  I'm not a big 
fan of very compact syntaxes with multiple ways to write the same thing 
because the regularity helps the first time reader.

	Andy

(/me runs and hides from Ivan and Sandro. :-)

>
> best,
>
> nathan

Received on Tuesday, 1 March 2011 17:57:46 UTC