Language Maps [was RE: [uf-discuss] Microformats vs XML]

[Joe Andrieu]

> From: Tantek Çelik  Sunday, April 30, 2006 6:56 PM
> To: microformats-discuss
> Subject: Re: [uf-discuss] Microformats vs XML
> 
> On 4/30/06 6:20 PM, "Karl Dubost" <karl at w3.org> wrote:
> > And your page has class names in English when you are using another
> > language. -1

[snip]
> Thus with microformats, you may use both the standard microformat class
> names, AND class names in your own non-English language if you wish:
> 
> <span class="family-name soyad">Çelik</span>
> 
> ("soyad" is Turkish for family-name)
> 
> Whereas with POX markup standards, you are relegated to only using the
> element names from the spec.
> 
> <family-name>Çelik</family-name>

The current microformat model is certainly better than POX, but I think it
still leaves something to be desired. This approach still requires that
everyone uses the Microformats Approved(r) Anglo-biased namespace, even if
they get to add their own term to the class.  

To the extent we can enable other peoples and languages to "own"
Microformats and participate as first class citizens, I suggest it would be
a Good Thing(tm). Couldn't we allow a mapping of any microformat into any
language?  This seems to be a simple solution for both humans and computers.

If we utilize the microformat attribute I mentioned in my previous email (or
simply standardize on a profile mechanism), we could see something like
this:

<a class="hcarte"
microformat="http://www.microformats.org.fr/hcarte-profile">

And in the hcarte-profile, we see something like this (again, apologies for
any technical errors):

<a rel="reference" href="http://www.microformats.org/wiki/hcard-profile">
Carte de langue pour hCarte à hCard</a>

<class = "profile">
  <dl>
   <dt id='nom-et-prenoms' ref='fn'>nom-et-prenoms</dt>
    <dd>Le nom et prenoms</dd>
   <dt id='donne-le-nom' ref='given-name'>donne-le-nom</dt>
    <dd>Donne le nom</dd>
   <dt id='nom-de-famille' ref='family-name'>nom-de-famille</dt>
    <dd>Nom de famille</dd>
  </dl>
</class>

My apologies for my French, but hopefully it gets the idea across.
Initially I wasn't sure if the hCard reference terms should be
human-readable, rather than attributes of the dictionary entry. If we are
going with the culturally sensitive approach, then I think the profile
should be 100% human readable /in the presenting language/.

Thus, if a French author/developer discovers the above mentioned
microformat, they may simply use it the same way the first user did:

<a class=hcarte
microformat=http://www.frenchmicroformats.org/hcarte-profile>

All without any requirement of seeing or using English except the one
reference to hCard in the title of the profile. (And technically that could
be cut out).  Plus, the elements of the microformat are now semantically
relevant to the human author. So, the French soccer-dad who wants to put his
daughter's soccer schedule online doesn't have to keep mentally translating
between the English hcard class names and his native language. This makes
Microformats much easier to use, especially in cultures and countries where
English is not quite the standardized "second tongue."

Once any microformat profile has been mapped to a language, it is easily
language-friendly to anyone using that language, assuming they can find the
mapping (another argument for a central registry).

I expect some might see this approach as offering the potential for chaos.
However, the profile would still be a 1:1 mapping to a well-understood
microformat. And if we have a reliable profile mechanism, the automated
discovery of the semantic translation would be straightforward. Hence,
humans get to use the language they want and computers get clean semantic
data.  Isn't that what is at the heart of the Microformats approach?

Does this make any sense?


-j


--
Joe Andrieu
joe at andrieu.net
+1 (805) 705-8651