[uf-discuss] hCard: url and tel

[Tantek Ç elik]

On 1/7/08 4:01 PM, "Katrina" <Kaz at t-tec.com.au> wrote:

> Tantek Çelik wrote:
>> On 1/7/08 2:42 PM, "Andy Mabbett" <andy at pigsonthewing.org.uk> wrote:
>> 
>>> In message <C3A7E2D2.9A3C3%tantek at cs.stanford.edu>, Tantek Çelik
>>> <tantek at cs.stanford.edu> writes
>>> 
>>>>> In any case, how is that different from:
>>>>> 
>>>>>       <abbr class="dtstart" title="2008-01-07">7 Jan</abbr>
>>>>> 
>>>>> where "2008" is "hidden"?
>>>> title attribute is displayed in tool-tips
>>> in some, but far from all, browsers.
>>> 
>>>       <http://www.w3.org/TR/html401/struct/global.html#h-7.4.3>
>>> 
>>>       Values of the title attribute may be rendered by user agents in
>>>       a variety of ways.
>>> 
>>> 
>> 
>> Data in the class attribute is a known anti-pattern.  Not a new issue.
>> 
> 
> I admit I am new to microformats, however, I have always understood
> class attributes to hold a type of data: meta-data. They describe what
> sort of data that particular element holds.

Hi Katrina,

"what sort of data" is what a "type" is, and a type is very distinct from
data itself.

E.g. integer is a type. -1,0,1 are examples of data of that type.


> So if you have a list of
> books, instead of giving it a class attribute of red or blue, it should
> be labeled "books".
> 
> Or more pertinent to the microformats, class="vcard". vcard is meta-data
> saying that this particular element holds a vcard.

It is a very specific kind of meta-data that is a type.

Similar to how the <p> tag is saying this particular element holds a
paragraph.



> Reading this: 
> http://microformats.org/wiki/anti-patterns#data_in_class_attributes
> 
> I just don't seem to understand how the Microformats community decides
> what sort of meta-data is acceptable and what others aren't?

In general, the term meta-data tends to be more confusing than illuminating,
it means too many different things to different people, and thus we try to
avoid its use in discussions.

microformats simply extend the typing/schema information that HTML itself
has.

Where HTML has a limited vocabulary to markup what data/content is
paragraphs, blockquotes etc.

microformats extends that limited vocabulary, also in a limited way, to
markup what data/content is people, organizations etc.


> I also thought that Microformats were to take human data and translate
> that into machine-readable.

Not quite.  Mostly microformats just markup existing data in the page so
that machines can find it and know what type of data it is.  In some
instances a machine-readable instance of the data is necessary, but those
are both the minority and minimized if at all possible.


> In order to do so, context needs to be
> translated to make it machine readable.
> 
> If you come across a business selling something, as a human, you can
> determine that it is work related, and it doesn't need to be labeled
> 'work'. However, a machine cannot make that distinction and needs to be
> explicitly told. How does the Microformats community then allow people
> normal contextual communication but also specifies the contextual data
> for machines? 

The broader problem of arbitrarily extracting meaning from any human prose
is an Artificial Intelligence problem far beyond that of microformats, and
is deliberately not a goal. Hence why microformats focus on real world
examples on the web that are also *common*.

> Surely the way to do that is through meta-data?

See above.  The term meta-data is too heavily overloaded to be really useful
in a discussion.

Thanks,

Tantek