From: Rob Cameron <cameron@cs...> - 2001-06-19 15:28:53
What you essentially want, I think, is a solution to the "appropriate
copy" problem for books. When a user clicks on a book link, they
go to the library of their choice.
One issue with your proposed solution is the reliance on a global
server to interact with every click in redirecting through a cookie.
Another is that the form of the URL is awkward.
Here is what we have managed to implement in our research program
with "bibliographic protocol" to address this very problem.
(1) You create links to books with the "bibp:" scheme name
instead of "http:" and using the book's ISBN, e.g.,
bibp:ISBN/0-262-02473-X is the URI that links to "From Gutenberg
to the global information infrastructure by Borgmann"
(2) You enable all of the bibp: links in a document by pasting
a short bibp resolver into your HTML document. This resolver
is carefully written to enable bibp link service on all
Bibp links are also implemented natively within the latest
(3) To serve the users in a particular domain, say stateu.edu,
a library merely needs to implement a server at bibhost.stateu.edu.
The servers can do web-z39.50 gateways for example, using
(4) The bibp resolver checks for service at http://bibhost/. If the
local service is found, it is used. Otherwise, a known global
server is checked. For users at stateu.edu, http://bibhost/
translates to http://bibhost.stateu.edu/ using the built-in
relative domain name feature of DNS.
(5) We are developing a prototype for a credible global bibp:ISBN
server, see http://usin.org/ for examples. Basically, we
have developed ZI-Bot, a Z39.50-ISN robot that harvests pages
in the space of pages returned by ISN queries on known Z39.50
servers. In this way, the global server is populated with
the best data available from public servers; a sort of virtual
(6) Although libraries can install bibhost servers for default
service for a whole domain (users *never* need specify the
default server), it is possible and quite easy for to users
to override and choose their own favorite library or
bibliographic service provider: all that is needed is a one-line
entry in the "hosts" file on your computer giving the IP address
of the "bibhost" you want to use.
At 08:28 AM 6/19/01 -0700, you wrote:
>(1) You create links to books with the "bibp:" scheme name=20
> instead of "http:" and using the book's ISBN, e.g.,
> bibp:ISBN/0-262-02473-X is the URI that links to "From Gutenberg
> to the global information infrastructure by Borgmann"
Sounds like a great idea!
I'm not sure I'm getting this right, but what if you want to make a
reference to a book as an abstract entity, or to the work of art as such?
As far as I can remember from library-school, ISBNs refer to a specific
edition of a book, i.e. there are different ISBNs for the paperback- and
the hardbound editions of the same work.
Librarian & webmaster
Bod=F8 Regional University
From: Rob Cameron <cameron@cs...> - 2001-06-20 17:54:21
> >(1) You create links to books with the "bibp:" scheme name
> > instead of "http:" and using the book's ISBN, e.g.,
> > bibp:ISBN/0-262-02473-X is the URI that links to "From Gutenberg
> > to the global information infrastructure by Borgmann"
> Sounds like a great idea!
> I'm not sure I'm getting this right, but what if you want to make a
> reference to a book as an abstract entity, or to the work of art as such?
> As far as I can remember from library-school, ISBNs refer to a specific
> edition of a book, i.e. there are different ISBNs for the paperback- and
> the hardbound editions of the same work.
This is a good point. However, using an ISBN is only slightly more
specific than the traditional academic citation; in the latter,
edition, publisher and date are routinely specified, although
format is not.
Ultimately, our goal is to address this from the server side: when
you click on a bibp:ISBN link, you would see information not only
about the specific edition and format cited, but also about other
versions. We can get some of this information from typical MARC
records now, but more work is needed.
Another approach would be to use search-based or metadata-based
linking as in OpenURL. This could allow for more abstract links,
at the expense of potential ambiguity. Although our research
is concerned with the identifier-based approach, there is no reason
why our "appropriate copy" technology could not accomodate both