Re: One more time: Words have meaning

At 3:48 PM +0000 9/2/02, Ian Hickson wrote:


>I am talking about a UA which complies to current W3C, IETF, ECMA and ISO
>standards. Not a UA which requires AI technology way beyond our current
>abilities.

It doesn't require AI technology way beyond our current abilities. It 
requires some simple heuristics. If you want to go the AI route (and 
you don't have to), this is well within the range of current expert 
systems. It absolutely does not require strong AI that duplicates a 
human brain.

>
>>  You claim, "the class attribute is exactly the same as XML tag names
>>  in unknown namespaces -- UAs can have no clue what they mean."
>
>I don't "claim" that -- it is DEFINED that way.

So what? Is there really anything in the HTML spec that says UAs 
aren't allowed to assign their own uses to the class attribute? If 
so, what's the point of the thing?

However, even if that's true of HTML, this is demonstrably not true 
of XML. The <SINGER>Madonna</SINGER> element can have whatever 
semantics the local process chooses to apply to it. That certainly 
includes recognizing the English word SINGER and using it to 
distinguish between Madonna Louise Ciccone and Madonna, the mother of 
Jesus.

>
>>  For a simple example, that's clearly implementable today, imagine a
>>  user agent that allows the user to select an element such as
>>  <SINGER>Madonna</SINGER> and request more information on that [...]
>>  the UA could search for Madonna within the Arts:  Music: Vocal:
>>  Singers category instead of the entire database.
>
>And how does your UA know that "singer" refers to what we call music, and
>not to a device intended to lightly burn hair?

Probably because the sword singer is almost always used with the 
first meaning and almost never with the second. Perfect? No. Useful? 
Yes.

Even more likely, it simply recognizes that there's a category at 
DMOZ that contains the word "Singer" and it applies that without any 
real understanding of what that category means. Again, not perfect, 
but useful.

>>  You could not do this with merely <span>Madonna</span> because the
>>  span element does not contain as much information as the SINGER
>>  element.
>
>Surely by your reasoning the UA could look up Madonna in google and work
>out that it is a singer.

Maybe in this example, because Madonna is so popular that Madonna 
Louise Ciccone  is the subject of the first five web pages on the 
subject. On the other hand, <SINGER>Susan Tomes</SINGER> is a lot 
tougher to handle without the extra hint. More information is better. 
There's more information in "<SINGER>Susan Tomes</SINGER>" than in 
"Susan Tomes".
-- 

+-----------------------+------------------------+-------------------+
| Elliotte Rusty Harold | elharo@metalab.unc.edu | Writer/Programmer |
+-----------------------+------------------------+-------------------+
|          XML in a  Nutshell, 2nd Edition (O'Reilly, 2002)          |
|              http://www.cafeconleche.org/books/xian2/              |
|  http://www.amazon.com/exec/obidos/ISBN%3D0596002920/cafeaulaitA/  |
+----------------------------------+---------------------------------+
|  Read Cafe au Lait for Java News:  http://www.cafeaulait.org/      |
|  Read Cafe con Leche for XML News: http://www.cafeconleche.org/    |
+----------------------------------+---------------------------------+

Received on Monday, 2 September 2002 14:54:04 UTC