[XML4LIB] Re: The impact of XML/RDF on digital libraries

Jerome McDonough jerome.mcdonough at nyu.edu
Wed Feb 20 12:33:24 EST 2002


At 08:58 AM 2/20/2002 -0800, Eric Miller wrote:
>and from the recent (excellent) post from Andrea Giuliano "Re: Databases or
>XML" interestingly provides a similar relation:
>
>[[
>as someone already pointed out on the list, XML is a good choice to
>exchange data between application, because of the tree structure of an
>XML document, a very versatile data structure. But, from a mathematical
>point of view, relational databases are graphs, and trees are just a
>very particular kind of graphs. So every tree is itself a graph but, on
>the other hand, not evey graph can be arranged as a tree. So you can run
>across a database which cannot be arranged as a tree, whatever you do.
>]]
>
>RDF is based on a graph model; one that uses URIs for allowing the entities
>and relationships to be first class objects.  One way of viewing this is
>that RDF provides a simple model that enables effective merging and
>integration of data; "joining" different parts of the web in a similar
>fashion to joining tables in a relational database.  Hmm... seems very
>relevant to the library community to me :)

I believe Andrea's article also pointed out that the addition of ID/IDREF 
attributes
in XML provided a means of encoding complex graphs (like relational databases)
that can't be modelled in a pure tree.  So, this leads to the question I've 
always
had for RDF; what exactly does it give me  that I should care about that I
don't already have in plain XML?  If I'm really concerned about semantic 
interoperability
of diverse digital library systems (and I am) then RDF isn't sufficient in 
and of itself.  I need
a far more detailed set of agreements between myself and other people I want
to interoperate with as to what the exact set of elements we want to use 
is, what
their relationship to each other is, and how to put data inside those 
tags.  Sure,
I can use RDF as an encoding mechanism once I've reach those agreements;
but I can also develop my own, simpler XML syntax.  So, what good is RDF 
doing me
here?  The only answer I've gotten when I've pressed that issue is that if 
everyone
uses RDF, we'll be able to create the semantic web and all of these nifty AI
capabilities on the net, and frankly, that strikes as incredibly unlikely.

Given Eric's position, I'd be interested in his thoughts on this.  As I 
mentioned
to Simon off-line, standards get adopted because people think there's something
in it for them in the long term.  I'm heading a digital library unit at a 
major university.
What exactly is in RDF for me that I can't get through simpler means?

Jerome McDonough
Digital Library Development Team Leader
Elmer Bobst Library, New York University
70 Washington Square South, 8th Floor
New York, NY 10012
jerome.mcdonough at nyu.edu
(212) 998-2425



More information about the xml4lib mailing list