Jules,
> CIMI (http://www.cimi.org/) recently released Guide to Best
> Practice: Dublin Core and I wonder if we should apply Dublin Core on
> our website.
Indeed they have, and very good it is, too...! ;-)
As for the question of embedding Dublin Core in the HTML of your web
pages, that is actually quite a difficult one to answer, as it depends
what you're doing it for...
> So I address to the list the following questions :
>
> 1 - Is Dublin Core recognized and processed by the main web
> search engines such as Altavista, HotBot, Lycos, etc ?
No, it's not. None of the big web search engines will do anything
meaningful with Dublin Core that they find in an HTML web page. It has
been suggested that they deliberately don't make use of things like
Dublin Core because it is not in their best interests to do so... Search
Engines make money from advertising. The more pages of hits you send to
a user, the more adverts they see, and the more money you make. If you
use Dublin Core to reduce the number of hits, you send less pages to the
user, they see less adverts, and you get less money. Cynical, eh?
It seems likely that if *one* big search engine makes the plunge, it
will attract such a huge market share as users flock to the only
'reliable' search engine that the others will be forced to follow. Which
Search Engine is brave enough to make that leap, I wonder?
All is not lost, though, as a lot of the search engines you can set up
yourself to manage data on your intranet can make use of
user-configurable elements. There's no reason why you couldn't set the
15 Dublin Core elements up in this way, and use your own search engine
to offer Dublin Core-enabled searches of your web pages to visitors, or
to improve searches for staff over your own intranet.
> 2 - Does this standard really help users to locate Web resources with
> accuracy with all search engines?
See above. Also, even if you are using a search engine on your intranet,
*and* it understands Dublin Core, you're *still* going to need other
things to facilitate recall. It's all very well, for example, to put
names into the Dublin Core Creator element, but without the control
offered by something like the Anglo-American Cataloguing Rules (AACR2)
you actually haven't improved recall very much... Recording a single
individual's name as 'Miller, A.P, 1971-', 'Miller, P', 'Miller, Paul',
'Paul Miller', etc. at different times in the same resource and then
expecting users and software to accurately retrieve records by all of
them and *know* they're the same person is just asking for trouble...
Dublin Core is one part of the solution. Authority Controls and good
cataloguing practice are another.
> 3 - Do you use Dublin Core? (If so, please give me URL of such pages)
Yup. A lot of the UKOLN pages include Dublin Core metadata (see
http://www.ukoln.ac.uk/, and pages below that). Most of the metadata is
generated by a tool called DC-Dot
(http://www.ukoln.ac.uk/metadata/dcdot/), which was written by Andy
Powell at UKOLN.
Dublin Core isn't just for embedding in HTML web pages, of course. The
Arts & Humanities Data Service (AHDS), for example, uses Dublin Core as
a 'lowest common denominator' between data provided by five service
providers covering archaeology, history, the performing arts, the visual
arts, and textual studies. See http://ahds.ac.uk:8080/ahds_live/ for an
example of the system at work.
Paul
-- dr. paul miller - interoperability focus - [log in to unmask] --
u. k. office for library and information networking (ukoln)
tel: +44 (0)1482 466890 mobile: +44 (0)410 481812
---------------------------- http://www.ukoln.ac.uk/interop-focus/ --
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|