Re: Meta - Resource-type tagging

Jon Knight (jon@net.lut.ac.uk)
Thu, 26 Sep 1996 10:50:37 +0100 (BST)


Date: Thu, 26 Sep 1996 10:50:37 +0100 (BST)
From: Jon Knight <jon@net.lut.ac.uk>
To: Peter Flynn <pflynn@curia.ucc.ie>
cc: www-html@w3.org
Subject: Re: Meta - Resource-type tagging
In-Reply-To: <199609260749.IAA17595@curia.ucc.ie>
Message-ID: <Pine.SUN.3.95.960926102321.16187E-100000@weeble.lut.ac.uk>

On 26 Sep 1996, Peter Flynn wrote:

> > <META NAME="DC.author" CONTENT="(type=name) Jon Knight">
> > <META NAME="DC.title" CONTENT=" My New Ode">
> > <META NAME="DC.objecttype" CONTENT="(scheme=DCObjects) Poem">
> > <META NAME="DC.identifier" CONTENT="(scheme=URL) http://www.me.com/">
> 
> While this is perfectly unexceptionable, I still think the LINK
> element makes a better vehicle for the identification of links to
> other retrieveable objects. 
> Can you explain what advantages your last line has over the LINK 
> equivalent?

Well actually we propose using LINK as well to point to the human readable
descriptions of the DC elements and sub-element schemes.  See the DC
metadata embedded in <URL:http://www.roads.lut.ac.uk/> for an example of
this.  The reason we didn't use LINK to hold the DC.identifier is that we
wanted a nice, clean way of holding all of Dublin Core in more or less the
same format.  Stripping DC.identifier out into a LINK whilst all the
others lived in META tags would be a bit yucky.  Of course you, or your
browser will have the URL already as it would be needed to get the
document in the first place, so the point is pretty moot in this case. 
The DC metadata embedded in HTML is intended for describing just that HTML
document.  There are other formats for transporting DC for multiple
documents, etc, such as the SGML DTD that Lou Burnard, Eric Miller, Liam
Quin, and C.M. Sperberg-McQueen put together (see
<URL:http://www.uic.edu/~cmsmcq/tech/metadata.syntax.html>). 

> What software implements DC metadata?

Well there's several people writing code that lets you insert DC metadata
into web pages (including a rather neat CGI script that I just got to see
yesterday) but they're not quite ready for prime time yet. I believe that
SoftQuad were planning on implementing the above form of metadata
insertion into one of the forthcoming versions of Hot Metal (their HTML
editor).  The ROADS Project (the UK Electronic Libraries Programme Project
I work for) is building a harvesting module to allow us to collect DC
metadata from suitable HTML documents.  I understand that the original
proposal for the basic format that we've slotted DC metadata into was made
at the W3C Distributed Index and Searching workshop with Lycos, Microsoft,
Verity and WebCrawler people present, so I would expect that they'll be
looking to implement code to handle it as well.

Also there's been alot of support for the concept of Dublin Core metadata
from the library community, with work having been done on mapping DC
elements to USMARC (see MARBI Discussion Paper No. 86 at
<URL:http://www.uic.edu/~cmsmcq/tech/metadata.syntax.html>).  So its a
happening scene.

Tatty bye,

Jim'll

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Jon "Jim'll" Knight, Researcher, Sysop and General Dogsbody, Dept. Computer
Studies, Loughborough University of Technology, Leics., ENGLAND.  LE11 3TU.
* I've found I now dream in Perl.  More worryingly, I enjoy those dreams. *