W3C home > Mailing lists > Public > public-esw-thes@w3.org > October 2009

Re: language tags in RDF/XML vs. Turtle [was: Re: flashminutes from the Ecoterm meeting]

From: Ivan Herman <ivan@w3.org>
Date: Mon, 12 Oct 2009 06:57:11 +0200
Message-ID: <4AD2B727.20506@w3.org>
To: Thomas Bandholtz <thomas.bandholtz@innoq.com>
CC: Antoine Isaac <aisaac@few.vu.nl>, Sřren Roug <soren.roug@eea.europa.eu>, public-esw-thes@w3.org, Stefan Jensen <Stefan.Jensen@eea.europa.eu>, dave@dajobe.org, timbl@w3.org


Thomas Bandholtz wrote:
> Hi Ivan,
> 
> your helpful response did not reach public-esw-thes@w3.org as it 
> inherited my typo in the CC line ... I corrected that
> 

:-) And I should have checked... Thanks!

> It is important for this list, as the subject originated from a SKOS 
> serialisation of GEMET by the European Environment Agency, and has 
> impact on a current application of SKOSXL ("UMTHES").
>

I hope when it is finalized we can give some more publicity to it!

[skip]

> You may be right. I can live with omitting XML default language tags on 
> the node (or even document) level and provide them with each literal 
> individually (though this does not exactly mean the same).
> 
> What still makes me wonder is the issue of equivalence of the two 
> serialisations (XML vs. Turtle) of the same graph.
> Referring to the originating issue of this thread (see at the bottom): 
> if EEA would express the same GEMET graph in Turtle, they would have to 
> change the language tag pattern just as Antoine has expected, right?
> 

In the case of Turtle, yes.

> To me this apears somehow puzzling, see:
> http://www.w3.org/TeamSubmission/turtle/:  /"Turtle/ language, an 
> alternative syntax to RDF/XML."
> http://www.w3.org/TR/owl2-primer/#OWL_Syntaxes: "The examples and the 
> sample ontology in the appendix can be viewed as any of the four 
> different syntaxes, and we provide both RDF/XML [RDF Syntax] and Turtle 
> [RDF Turtle Syntax] serializations for the RDF-based syntax."

The sentence you quote from turtle is a little bit misleading indeed. 
Turtle is an alternative too RDF/XML insofar as its main goal is to 
encode RDF triples. It is also an alternative syntaxt to N-triples...

I am still not sure we have a problem. There are different syntaxes, 
each have their pros and cons; the question whether you can encode all 
valid RDF graphs in all of them and the answer is clearly yes on that. 
Beyond that, each syntax has its pros and cons.

Although I tend to write down things in Turtle, one haas to see that 
there _are_ some things in RDF/XML that are nice and it becomes more 
verbose in Turtle. xml:lang is one; another example might be the setting 
on XML node level the namespaces. This is, theoretically possible in 
RDF/XML and there is no analogy for that in turtle. I am sure there are 
other examples.

That being said: parsing RDF/XML is much more complex. It requires a 
full XML parser. Of course, one can get those off the shelf these days, 
but it is nevertheless it shows the complexity. The goal for turtle is 
to make it much simpler... Also, for a casual reader it might become 
more complex to understand RDF/XML, not the least because of all the 
complexities of XML. Turtle is much  more readable in my view...

(B.t.w.: I have not checked the historical records on all that. Turtle 
originates from TimBL's N3; it is N3 minus some syntax tricks. I wonder 
whether N3 was not created before the first version of RDF/XML was 
defined...)

 >
 > May be this should at least be clarified in
 > http://www.w3.org/TeamSubmission/turtle/
 > What do you think?
 >

I would leave the answer to the authors of the document...

Cheers

Ivan

> Kind regards,
> Thomas
> 
>>
>> Cheers
>>
>> Ivan
>>
>>
>>
>>
>>
>>
>>> Kind regards,
>>> Thomas
>>>
>>> Antoine Isaac schrieb:
>>>> Hi Sřren,
>>>>
>>>> Really sorry for the mistake! I actually did not know that XML rule,
>>>> and had *never* seen it applied in the RDF/XML that I've browsed so 
>>>> far.
>>>> But it loads perfectly in an RDF repository, it seems. I'm even more
>>>> impressed by your work, now :-)
>>>>
>>>> Best,
>>>>
>>>> Antoine
>>>>
>>>> PS: That being said, this story is yet another argument in favour on
>>>> non-XML syntaxes like Turtle; Thomas will appreciate!
>>>>
>>>>> Hello Antoine,
>>>>>
>>>>> Actually, that's not a bug. The xml:lang is set on the parent element
>>>>> - i.e. the <rdf:RDF> element. The W3C specification for xml:lang
>>>>> (http://www.w3.org/TR/REC-xml/#sec-lang-tag) states that the
>>>>> attribute applies to subelements as well.
>>>>>
>>>>> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
>>>>>             xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#"
>>>>>             xmlns:skos="http://www.w3.org/2004/02/skos/core#"
>>>>>            
>>>>> xmlns:gemet="http://www.eionet.europa.eu/gemet/2004/06/gemet-schema.rdf#" 
>>>>>
>>>>>
>>>>>             xml:lang="de">
>>>>> <rdf:Description
>>>>> rdf:about="http://www.eionet.europa.eu/gemet/concept/7">
>>>>> <skos:prefLabel>Altstandort</skos:prefLabel>
>>>>> </rdf:Description>
>>>>>
>>>>> Best regards,
>>>>> Sřren Roug
>>>>>
>>>>>
>>>>> On 09-10-2009 17:08, Antoine Isaac wrote:
>>>>>> Dear Stefan, Soren,
>>>>>>
>>>>>> Thanks for establishing the connection!
>>>>>>
>>>>>> First, I'd like to repeat that I was very impressed to see that
>>>>>> Gemet was already available as linked data. That's great work!
>>>>>>
>>>>>> The "bug" I have spotted is quite simple: at [1] there is the option
>>>>>> to get SKOS data for each specific language addressed by GEMET. For
>>>>>> instance, you can get the labels of all concepts, as in the
>>>>>> following RDF/XML snippet (for English):
>>>>>>
>>>>>> <rdf:Description
>>>>>> rdf:about="http://www.eionet.europa.eu/gemet/concept/8">
>>>>>> <skos:prefLabel>abandoned vehicle</skos:prefLabel>
>>>>>>
>>>>>> or that one (for French):
>>>>>>
>>>>>> <rdf:Description
>>>>>> rdf:about="http://www.eionet.europa.eu/gemet/concept/8">
>>>>>> <skos:prefLabel>véhicule abandonné</skos:prefLabel>
>>>>>>
>>>>>> The problem is that in those versions, the language tags for the
>>>>>> labels (and for the definitions) are missing. If someone decides to
>>>>>> load these downloaded files into a same RDF repository, there will
>>>>>> be therefore problems, especially with respect to having several
>>>>>> preferred label with the same (lack of) language tag. This should be
>>>>>> quite easy to solve, though, as your linked data version at [2]
>>>>>> returns the perfect data when asked for rdf+xml:
>>>>>>
>>>>>> <skos:prefLabel xml:lang="en">abandoned vehicle</skos:prefLabel>
>>>>>> <skos:prefLabel xml:lang="fr">véhicule abandonné</skos:prefLabel>
>>>>>>
>>>>>> I hope this is useful feedback for you!
>>>>>>
>>>>>> Best regards,
>>>>>>
>>>>>> Antoine
>>>>>>
>>>>>> [1] http://www.eionet.europa.eu/gemet/rdf
>>>>>> [2] http://www.eionet.europa.eu/gemet/concept/8
>>>>>>
>>>>>>       
>>>>>
>>>>
>>>
>>>
>>
> 
> 
> -- 
> Thomas Bandholtz, thomas.bandholtz@innoq.com, http://www.innoq.com 
> innoQ Deutschland GmbH, Halskestr. 17, D-40880 Ratingen, Germany
> Phone: +49 228 9288490 Mobile: +49 178 4049387 Fax: +49 228 9288491
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf


Received on Monday, 12 October 2009 04:54:06 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:39:04 GMT