[eml-dev] xml:lang attribute for title in EML 2.1.0

Mark Servilla servilla at lternet.edu
Thu Sep 16 19:18:38 PDT 2010


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

I agree with the changes proposed by Ben in his 7/30/10 Jul 30, 2010 email -
that is, the "mixed content" approach.  It is not entirely clear what the
long-term implications of the changes will be for parsing and validation, but
the short-term approach seems manageable (albeit, with effort from some warm
body).  Internationalization of EML is a necessary step as we move research and
its associated data/metadata to a global level.

I personally do not see the alternatives as being viable long-term solutions.
Multiple documents of differing languages will ultimately be too cumbersome and,
likely, not kept synchronized.  Introducing another inline content attribute
(i.e., <title>[Language:En]Snow cover...) only adds yet more syntactical parsing
issues.  The use of the xml:lang attribute is, at least, a recognized and
standard approach in many systems.

I appreciate and thank Ben and others for their efforts in this matter.

Sincerely,
Mark

On 9/16/10 Sep 16, 2010 1:56 PM, Matt Jones wrote:
> The solution that Ben proposed is meant to address the requirements that arose
> from the iLTER Lake Taihu meeting for providing core metadata in multiple
> languages.  These recommendations then were also at the core of the
> recommendations made to GBIF about which fields should contain English
> translations, but the set of fields differs slightly in the two recommendations.
>  Because many of these fields are not currently repeatable according to the EML
> 2.1 schema, we would need to, at a minumum, change cardinality rules to allow
> for each field to be included multiple times if the xml:lang tag were used to
> differentiate them (or for the approach Inigo points to).  As Ben points out, it
> would still be ambiguous as to whether the repeating fields represent different
> information, or the same information translated.  So his proposal is meant to
> explicitly flag translations as such within mixed content string fields, with
> the goal of doing so without breaking existing EML 2.1 compatibility and without
> having to change existing cardinality rules.
> 
> Ben's prior discussion on this highlighted the conflict with the NonEmptyString
> type that was introduced in EML 2.1, in that mixed content elements would not be
> validated and so the rules for NonEmptyString would not be enforced.  I think
> this would only be a small issue, and that the advantages in compatibility
> provided by using a mixed content model for language translations outweigh the
> loss of validation within our string types.  Either way, we would need to add
> the xml:lang attribute so that it can be used throughout EML, including in the
> translation elements that Ben proposed.
> 
> Are there any objections to moving forward with the schema changes to use a
> mixed content models for translations that Ben proposed in his earlier emails?
> 
> Matt
> 
> On Thu, Sep 16, 2010 at 11:35 AM, Inigo San Gil <isangil at canyon.lternet.edu
> <mailto:isangil at canyon.lternet.edu>> wrote:
> 
> 
>     We'll keep our eyes on the ball, then.
> 
>     Meanwhile others have adopted their own solution.
>     Here are two examples:
>     1) a site from Spain reports this implementation
> 
>     <title>[Language:En]Snow cover data provided by MODIS satellite imagery</title>
>     <title>[Language:Sp]Datos de innivaci&#243;n seg&#250;n im&#225;genes
>     MODIS</title>
> 
>     We thought that the use of the XML attribute "lang=en | sp"
>     was interesting -but, among other problems,  we would  have
>     gotten screwed by eml-dev eventual internationalization
>     implementation.  Call it luck, but you can bet the "eventual
>     eml-dev decision" would force us to re-code the EML
>     generation.
> 
>     2) From Taiwan, it is also a mix and match.  I had the
>     internationalization conversation years ago, when we set
>     harvesting into the NBII clearinghouse.  at the TFRI, we
>     found EML documents that have a hybrid of english and
>     chinese, with no sign or whatsoever of the language used.
>     We had to devise a mechanism to detect language.  We
>     simply did not harvest those docs whose critical content
>     was not translated in English.
> 
>     ILTER discussed (two years ago?) some guidelines on
>     how the different countries were going to deal with the
>     tower of Babel problem.  May be you can look into those
>     if you feel curious, but if I recall correctly, it went along
>     the lines of encoding the metadata in the native language,
>     and produce some discovery-level EML in English. This
>     strategy would create two EMLs per EML..
> 
>     Sparks or not, I still have to recommend the EML users
>     to implement some solution. Im inclined to suggest  that
>     such solution 1) does not break the current EML rules.
>     2) The solution should allow for easy language detection.
>     Spain's case fits here, for example.
> 
>     Cheers, inigo
> 
> 
> 
> 
> 
>     On 9/16/2010 10:56 AM, ben leinfelder wrote:
> 
>         Hi Markus,
>         I'm afraid your findings are accurate with respect to the xml:lang
>         attribute in the<title>  element (or any "NonEmptyStringType" element).
>         In the course of my experimentation with allowing backwards-compatible
>         internationalization with a new EML version (2.1.1) I did have to
>         include the "http://www.w3.org/XML/1998/namespace" namespace just as you
>         did and also declare the xml:lang attribute in elements where I wanted
>         to employ it.
>         While certain EML elements are repeatable, it's not always clear what
>         the presence of multiple elements represent (are they translations in
>         different languages or are they alternate titles?). In order to clarify
>         this confusion and also allow multiple translations for non-repeatable
>         elements I proposed a solution for allowing mixed element content for
>         fields that should be internationalized. There's a fairly comprehensive
>         discussion of this approach in our eml-dev archives:
>         http://mercury.nceas.ucsb.edu/ecoinformatics/pipermail/eml-dev/2010-July/001828.html
>         I didn't get a lot of decisive feedback and so have not moved forward
>         with releasing an updated EML version. Hopefully this thread will again
>         set the ball rolling.
>         -ben
>         .nceas.ucsb.edu/ecoinformatics/mailman/listinfo/eml-dev
>         <http://nceas.ucsb.edu/ecoinformatics/mailman/listinfo/eml-dev>
> 
> 
>     _______________________________________________
>     Eml-dev mailing list
>     Eml-dev at ecoinformatics.org <mailto:Eml-dev at ecoinformatics.org>
>     http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/eml-dev
> 
> 
> 
> 
> _______________________________________________
> Eml-dev mailing list
> Eml-dev at ecoinformatics.org
> http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/eml-dev

- -- 
Mark Servilla, Ph.D.

LTER Network Office
Department of Biology
MSC 03 2020
1 University of New Mexico
Albuquerque, NM 87131-0001

servilla at LTERnet.edu
Office (505) 277-2619
Cell   (505) 453-8593
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkySz/4ACgkQqFW3+12RyXOEggCeLtSSf8r3pJty+lv06lk9uSVH
z0YAn1HQNykMFDCt8zIm02bwMv5iecng
=z21i
-----END PGP SIGNATURE-----


More information about the Eml-dev mailing list