Re: Sloppy inference rules

Le 01/11/2012 14:10, Markus Lanthaler a écrit :
> On Thursday, November 01, 2012 11:19 AM, Antoine Zimmermann wrote:
>
>
>>> Honestly it sounds a bit strange to me to simply accept that there is
>> a
>>> fundamental problem without trying to address it - especially
>> considering
>>> that the problem has been known since at least 2005 (2002?).
>>
>> What problem are you referring to? We know well since 2005, or maybe
>> earlier, that the rules are not complete. We will fix it.
>
> Great!
>
>
>> The bnode-in-predicate and literal-in-subject issues are not problems,
>> they are simply decisions to restrict the language.
>
> Why are those restrictions there? I'm specifically talking about
> bnode-in-predicate.

The restrictions, as far as I can track it, is because originally, the 
RDF rec in 1999 was closely tied to its serialisation in RDF/XML where 
it is not possible to have bnodes as predicates. Extending RDF/XML in 
such a way that it would allow bnodes in predicate position is 
difficult, so it stayed in the 2004 spec. For literals in subject 
position, it is similarly not possible to do it in RDF/XML but it would 
be much easier to extend it to do so. Yet, it was not done until the 
spec finally became Rec in 2004, but there was indication that it would 
be certainly fixed in a future version (even SPARQL 1.0 says that this 
restriction is likely to be removed, thus literals are allowed in 
subject in SPARQL syntax). Nonetheless, implementations took advantage 
of the restriction, in order to process subjects faster, and those 
restriction are now present all over the place in many applications. For 
these reasons, our WG has chosen not to change the restriction.

[Personal remark: in spite of these arguments, I believe that allowing 
literals in subject position is a trivial fix and I'm vehemently in 
favour of it for any future spec. I am more mitigated regarding bnodes 
in predicate position.]


>> We have accepted at the F2F that JSON-LD is able to do more than just
>> RDF graphs and as far as the persons present at the meeting are
>> concerned, it seemed to be ok for the WG members. So I don't see a big
>> issue here.
>
> OK, I didn't have that impression and also David's comment suggests that
> that's not really the case.

Basically, whenever an RDF processor use JSON-LD, it can simply ignore 
the stuff that are not kosher RDF. Again, if I may give my personal 
opinion, I would prefer that JSON-LD be a pure RDF serialisation, but I 
can easily live with the minor extra stuff that it allows.


--AZ

>
>> It is also consistent with my suggestion that JSON-LD
>> state in their spec that there may be valid JSON syntax that is not
>> valid JSON-LD.
>>
>> There needs to be a "fix" here, but it is an easy one not a major
>> rewrite.
>
> .. but maybe there's also just a misunderstanding on my side.
>
>
> Regards,
> Markus
>
>
> --
> Markus Lanthaler
> @markuslanthaler
>
>


-- 
Antoine Zimmermann
ISCOD / LSTI - Institut Henri Fayol
École Nationale Supérieure des Mines de Saint-Étienne
158 cours Fauriel
42023 Saint-Étienne Cedex 2
France
Tél:+33(0)4 77 42 66 03
Fax:+33(0)4 77 42 66 66
http://zimmer.aprilfoolsreview.com/

Received on Thursday, 1 November 2012 15:11:19 UTC