Databases or XML

Alan Kent ajk at
Sun Mar 17 20:41:46 EST 2002


I am new to this list (so appologies in advance if I am off topic or breach
any list nettique), but noticed in the recent thread on "Databases or XML".

I just thought I would mention a few library-related activities going
on around the place that I know of that involve XML. Might be
a little off track from the original posting, but I thought might be
of interest anyway. Sorry if these are old hat.

I am the Technical Director/Chief Technology Office for a commercial
product which is a Z39.50 database server with native XML (and SGML)
support. But I will try to restrain to technology topics, not product
topics! :-)

Some areas I know of:

In the Open Archives Initiative Protocol for Metadata Harvesting (OAI-PMH),
XML is used both for protocol requests, and for delivering metadata. The
main metadata format used is Dublin Core based metadata in XML.
This is probably more digital library related. (

Within Z39.50 (library search and retrieval protocol) XML is an "official"
record syntax. That is, Z39.50 can be used to send records in XML.
It is slowly gaining usage, but does not seem to be used that much yet.

Work is also in progress defining a simplified explain service for
Z39.50 servers, allowing sites without a Explain capable Z39.50
service to define an XML description about their target (and have
someone else host the information for them). ("Explain" is a Z39.50
concept allowing a client to dynamically discover the capabilities
of a database server. The Z39.50 protocol defined a complex form
of Explain, but not many vendors implemented it. The simplified Explain
is based on a single XML document describing the main points about
a Z39.50 database.)

Work is also in progress for defining a SOAP API under the ZiNG
(Z39.50 International: Next Generation) framework. This will allow
a Web Services interface (SOAP/.NET/etc) interface to a Z39.50 database.
SOAP is XML based. Again, records are only planned to be supported in XML.

There is the Dublin Core/MARC "crosswalk" which gives some sample
rules for turning XML to and from MARC. Not much experience here,
but I have often thought about developing a Z39.50 server that
does its best to turn MARC and/or Dublin Core XML records into
the other format.

We do lots of projects outside the library community using Z39.50
where we map Z39.50 attribute sets (Bib-1 or customized) onto collections
of XML documents. This allows the document richness and semantics to
be maintained (by keeping the documents in XML), but allowing simple
access points for searching (Z39.50 USE attributes). (Actually, this
is our main business! We have done almost no work within the library
community :-)

Alan Kent (mailto:ajk at,
Postal: Multimedia Database Systems, RMIT, GPO Box 2476V, Melbourne 3001.
Where: RMIT MDS, Bld 91, Level 3, 110 Victoria St, Carlton 3053, VIC Australia.
Phone: +61 3 9925 4114  Reception: +61 3 9925 4099  Fax: +61 3 9925 4098 

More information about the xml4lib mailing list