[XML4LIB] Re: XML indexing/searching (was: XML publishing

Alan Kent ajk at mds.rmit.edu.au
Wed Apr 17 21:21:42 EDT 2002

On Tue, Apr 16, 2002 at 12:31:51PM -0700, David Dorman wrote:
> Several ILS vendors claim to be able to support the creation and 
> searching of XML-based metadata and link that metadata directly to 
> digital resources.  Is no one using their ILS system to create and 
> index metadata and link the metadata to their digital resources?
> David Dorman

I hope its not a breach of nettique to point you at TeraText (formerly
SIM). I am the CTO for the TeraText development group. TeraText is not
a ILS as such - but it is a high performance text database system with
native XML support. It also uses Z39.50 as its native query protocol,
hence the overlap with the ILS field. So this reply might be slightly
off target, but there were some other items in this thread talking
about other products, so I thought I would sneak in and wave our own flag.

As per scale, we frequently deal with 10 - 100 gb databases. Our largest
installation (with a lot of hardware behind it I will admit) has around
1 billion XML metadata records in a distributed configuration (our server
supports query referals to other servers allowing us to distribute.
The site loads and indexes interactively hundreds of records per second.
Simple queries across the full collection take around 2 seconds to respond.
Million record databases on a single machine are typically around the
20ms range to answer (asuming good hardware, primed disk caches etc).
I mention this only because some mail in this thread talked about the
lack of scalability of some XML solutions. We scale well, but probably
cost more (I am in the technical side, not the sales side, so don't
know how much other products cost - actually, I am not 100% sure what
our our product costs! ;-)

You may not have heard much of us yet as we only recently officially
launched with a new distributor in the US. We have not really been in
the library community much to date (libraries here in Australia don't
have much money! - and we don't have cataloging or anything like that,
just Z39.50 support for querying text collections). If you are interested,
feel free to visit www.teratext.com. I actually work for RMIT University
in Australia where the technology originated, under contract to a
spin off company www.inquirion.com. Our first commercial sale of
the product in Australia was back in 1994 (so we have been around
a while).

Sorry if this is too commercial for this list. I am happy discuss 
any details of the technical approaches we have used either on or
off the list as appropriate. We have had SGML support since our
first release, and XML support shortly after XML came into existance.
Most application we develop involve searching SGML and/or XML.

Oh, I have been personally (on a non-official basis) involved recently
in the OAI protocol, Z39.50 interoperability of Australian Z39.50 public
servers, and the Australian version of GILS (see my personal web page in
signature below if your are interested). I have also been recently
playing with support for spatial queries for the Z39.50 GEO profile.
I am happy to talk about any XML (or SGML) related issues applicable
to libraries in those areas too.

Alan Kent (mailto:ajk at mds.rmit.edu.au, http://www.mds.rmit.edu.au/~ajk/)
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