Re: Meta Tag [and Knowledge Representation]

Robert Hazeltine (rhazltin@zeppo.nepean.uws.edu.au)
Mon, 20 Nov 1995 11:42:54 +1000 (EET)


Date: Mon, 20 Nov 1995 11:42:54 +1000 (EET)
From: Robert Hazeltine <rhazltin@zeppo.nepean.uws.edu.au>
To: "Daniel W. Connolly" <connolly@beach.w3.org>
Cc: Davide Musella <davidmsl@anti.tesi.dsi.unimi.it>, kmm@w3.org, dsr@w3.org,
Subject: Re: Meta Tag [and Knowledge Representation]
In-Reply-To: <199511171610.LAA15706@beach.w3.org>
Message-Id: <Pine.A32.3.91.951120110518.10304E-100000@zeppo.nepean.uws.edu.au>

Dan

On Fri, 17 Nov 1995, Daniel W. Connolly wrote:

> 
> I'd like to point out that this thread has escaped the bounds of
> HTML, and has become a discussion of knowledge representation in
> general.

Maybe, but there is nothing like a little debate to sharpen up ideas. No?

> It happens quite a bit. Much of the value of HTML is its ability to
> capture and represent knowledge in an informal or semi-formal way.
> But folks often run into the boundaries of its expressive capability,
> and look for ways out.
> 
> The META tag is intended for experiments on the edge of HTML's
> expressive capability. I haven't seen "the answer" to this problem
> that I'd like to standardize.

I think that this is begging the question.  The original poster has 
identified a serious deficiency in my view which is the lack of 
meta-information, whether or not the present or future specs deal with 
the problem with these elements.

If I can again re-iterate the problem:  applications can be identified 
(even if not always optimally) through the use of a file name and a 
README which contains information about authorship, versions and 
amendment dates etc sometimes as brief as a Revision Control System 
statement.

At least people in the know can, for example, identify the distribution 
copy of the netscape browser; others are going to pick it up at they go;
but it is identifable.

On the other hand, there is no equivalent for text material (HTTP aside).
Individual text files are the equivalent of a chapter, section or 
whatever torn from a book so that there is no cover or front page to 
identify it.  Combine that with duplicate names and you might begin to 
understand the problems of a mirror site that wants to archive text, 
even before we get to the location (URCs, etc), catalogue and permanency of 
material.

> The point of my message is to encourage those of you who are seriously
> interested in this topic to become familiar with the existing work in
> the field: both the traditional library science cataloging (sp?)
> stuff, and the more wigged-out AI knowledge representation stuff.
> 
> Reading the background literature will show just how interesting
> and complex this problem is. Hmm... maybe that's not 100% a good
> thing: the web was largely built by folks who didn't know enough
> realize that it couldn't be done :-)

So you really don't mind if I don't realise why something can't be done 
about this problem  :)

> [bibliography omitted]

Rob...