Re: Use of dc:language in test subject?

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Shadi,

I thought this might be a bit outside the core focus of EARL.

We will probably use dc:language as an application specific extension to
earl:TestSubject, in a similar way as outlined below, since that at
least will not break anything.

We also plan to record http content-language for the web resource involved.

Regards,
Nils

Shadi Abou-Zahra wrote:
> If I understand correctly, you want to filter the language from the Web
> resource (the test subject) and duplicate this information within the
> EARL report to improve query performance.
> 
> Your initial question was whether the WG is considering adopting a
> property for this use case. The answer to this question is not at this
> time. However, I am curious if the WG participants see a need for such a
> property in the core EARL vocabulary.
> 
> Meanwhile, I think one way to do this would be to use the dc:language
> property within the test subject as you initially pointed out. Since
> this is redundant information, you will not break anything. At the same
> time, you can not expect other tools outside your application to process
> or provide this type of information. Here is a simple example for this
> application-specific extension:
> 
> <earl:subject>
>  <earl:TestSubject>
>    <!--// core earl:WebContent properties //-->
>    <dc:title xml:lang="en">Page 1</dc:title>
>    <dc:location rdf:resource="http://example.org/page1.html"/>
>    <dc:format rdf:parseType="Literal">text/HTML</dc:format>
>    <!--// application specific properties //-->
>    <dc:language rdf:parseType="Literal">fr</dc:format>
>  </earl:TestSubject>
> </earl:subject>
> 
> 
> Another way to approach this would be to rely on the content-language
> information provided by the server instead of processing the Web
> resource itself. This information could be easily recorded in the
> http:content-language property of the HTTPresponse in a Web Content
> class. I am not sure how consistent and accurate this information
> usually is though.
> 
> 
> Regards,
>  Shadi
> 
> 
> Nils Ulltveit-Moe wrote:
> Hi Shadi,
> 
> We can probably use the lang attribute to the earl:TestSubject.
> 
> The use case is that our pool of assessment modules parse the (X)HTML
> pages that we want to perform automatic assessments on. We would like
> harvest information about the content language of the page being
> assessed for statistical analysis in a data warehouse, and we would like
> to utilise the fact that the assessment module already had this
> information available.
> 
> Best regards,
> Nils
> 
> Shadi Abou-Zahra wrote:
>>>> Right. I bow before you humbly. ;)
>>>>
>>>> The correct way would be to do this in the dc:description of the
>>>> earl:TestSubject instance that is referenced by the earl:subject
>>>> property. Here is a better example that shows both methods:
>>>>
>>>> <earl:subject>
>>>>  <earl:TestSubject>
>>>>    <dc:title xml:lang="en">left nav bar</dc:title>
>>>>    <dc:description rdf:parseType="Literal"><![CDATA[<xhtml xml:lang="fr"
>>>> ...>...</xhtml>]]></dc:description>
>>>>  </earl:TestSubject>
>>>> </earl:subject>
>>>>
>>>>
>>>> Still, I do not see a use case for dc:language. Correct?
>>>>
>>>> Best,
>>>>  Shadi
>>>>
>>>>
>>>> Johannes Koch wrote:
>>>>> Shadi Abou-Zahra wrote:
>>>>>> What about using the xml:lang attribute on the RDF nodes (for example
>>>>>> <earl:subject xml:lang="fr" .../>)?
>>>>>>
>>>>>> Or even better, when possible, within the content itself. For example
>>>>>> <earl:subject><![CDATA[<xhtml xml:lang="fr"
>>>>>> ...>...</xhtml>]]></earl:subject>.
>>>>> The object for earl:subject is not defined to be a Literal. It must be
>>>>> an earl:TestSubject.
>>>>>
> 
>>
>>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFE983N+uAhaUHEvIoRAgQcAJ4y1Sxutv3zWdd7WVWUQePIaCK+KgCgr3Or
6kAJuEcWT+REMxcm0x05LZk=
=om9l
-----END PGP SIGNATURE-----

Received on Friday, 1 September 2006 06:06:35 UTC