Re: is/of feature of N3 in turtle/sparql?

Eric Prud'hommeaux wrote:
> On Sat, May 28, 2005 at 03:47:50PM -0400, Stan Devitt wrote:
> 
>>Just to be perfectly clear, 
>>
>>the key point I want to raise is that  at the moment depending on how "is" 
>>aliased, the statement  ":a is :b of :c."  may or may not
>>parse.  Similarly for   ":a is :b."    The "is" tokens in the two 
>>statements play entirely different roles. 
>>
>>I am commenting specfically on n3 at the moment and need to look at the 
>>use on SPARQL before commenting further, other
>>than indicating my preference to avoid applying aliasing at tokenization 
>>time.
> 
> 
> The relevent part of the SPARQL grammar is at
>   <http://www.w3.org/2001/sw/DataAccess/rq23/#rTriples1>
> 
> The grammar currently inludes:
>   . , ;
>     s p o . s p o2
>     s p o, o2
>     s p o; p o2
>   [] ()
>     [] p o
>     [ p o ]
>     s p ( i1 i2 )

and indeed the SPARQL grammar bears a strong similarity to n3.n3 :-), including 
the optionality of ; in  [ :p :q ; ], only requiring DOTs between triple 
patterns and using tail recursion.

> 
> My opinion as both an editor and working group member is that the
> turtle syntax is intuitive and attractive in its simplicity. We have
> postponed a number of issues for later versions of the language. I
> would postpone the expansion of the graph language to include more of
> n3 such as:
>   s >- p -> o # now  s = p => o  ?
>   o <- p -< s # now  o <= p = s  ?
>   o is p of s
>   has  # which i don't understand. where's the object?
>   this
>   =  # for owl:sameAs
>   is  # without the of

Not in n3.n3 ?

>   and the ever-controversial {}

Neither Turtle nor SPARQL have property paths either which is more of an omission.

:: n3.n3 ::
path cfg:mustBeOneSequence(
  ( node pathtail )
 ).

pathtail cfg:mustBeOneSequence(
  (  )
  ( "!" path )
  ( "^" path )
 ).
:: n3.n3 ::

which, in practice, would be useful (that is, I wanted them for something I was 
doing yesterday!)

 Andy

PS When might there be a Turtle note?

> 
> 
> 
>>Stan Devitt
>>05/28/2005 03:24 PM
>>
>>
>>        To:     Danny Ayers <danny.ayers@gmail.com>
>>        cc:     Dan Connolly <connolly@w3.org>, "Eric Prud'hommeaux" <eric@w3.org>, 
>>public-cwm-talk@w3.org, public-cwm-talk-request@w3.org
>>        Subject:        Re: is/of feature of N3 in turtle/sparql?
>>
>>There are two important things to note:
>>
>>1.  The current n3 grammar defines this four term structure in terms of 
>>the tokens @is and @of.  This re-interpretatation in terms of "is" and 
>>"of" relies on
>>aliasing.
>>
>>2.  Because  @is and @of participate in a very different grammatical 
>>structure from (say) simple name tokens this complicates the aliasing 
>>mechanism considerably.  There is now more than one possible grammatical 
>>role for the target of the alias so the actual tokenizing is changed by 
>>the aliasing. 
>>
>>I would be happier if either aliasing were not involved, or at least did 
>>not change the tokenizing.  Can someone comment on this? 
>>
>>Stan Devitt
>>Agfa 
>>
>>
>>
>>
>>
>>Danny Ayers <danny.ayers@gmail.com>
>>Sent by: public-cwm-talk-request@w3.org
>>05/28/2005 05:03 AM
>>Please respond to Danny Ayers
>>
>> 
>>        To:     Dan Connolly <connolly@w3.org>
>>        cc:     public-cwm-talk@w3.org, "Eric Prud'hommeaux" <eric@w3.org>, (bcc: Stan 
>>Devitt/AWKCT/CAN/AGFA/CA/BAYER)
>>        Subject:        Re: is/of feature of N3 in turtle/sparql?
>>
>>
>>
>>Dan, I've still not spent any proper time with cwm (for shame!), could
>>you please just clarify is/of:
>>
>>* It's a convenience rather than introducing anything new? 
>>
>>* Would this be correct -
>>
>>select ?x where
>>{
>>:EricDoc dc:creator ?x
>>}
>>
>>select ?x where
>>{
>>?x is dc:creator of :EricDoc
>>}
>>
>>would produce the same results?
>>
>>Cheers,
>>Danny.
>>
>>
>>On 5/28/05, Dan Connolly <connolly@w3.org> wrote:
>>
>>>I just realized the other day that SPARQL doesn't have is/of,
>>>nor does turtle...
>>>
>>>22:48:25 <DanC> ^sparqlquery select ?pg where { [ foaf:nick "DanC"] is
>>>cyc:socialParticipants of [ contact:homePage ?pg] }
>>>22:48:25 <julie> Bad query (SPARQL syntax error at 'i')
>>>22:48:38 <dajobe> that's not in sparql - is, of
>>>22:48:38 <DanC> oh frap. no is/of in SPARQL
>>>http://www.ilrt.bris.ac.uk/discovery/chatlogs/swig/2005-05-19.html#T22
>>>-48-25
>>>
>>>Hmm.
>>>
>>>I'm hoping the WG will close all the remaining issues, including
>>>http://www.w3.org/2001/sw/DataAccess/issues#punctuationSyntax ,
>>>in the next week or so. In particular, punctuiationSyntax is pretty
>>>much all set to go this Tuesday. So it would be good to get feedback
>>>on this right away. In particular, if you want is/of in, send
>>>something to public-rdf-dawg-comments@w3.org .
>>>
>>>refs:
>>>
>>>N3 primer, on is/of etc. http://www.w3.org/2000/10/swap/Primer
>>>SPARQL http://www.w3.org/TR/rdf-sparql-query/
>>>turtle http://www.ilrt.bris.ac.uk/discovery/2004/01/turtle/
>>>
>>>
>>
>>
>

Received on Tuesday, 31 May 2005 09:27:19 UTC