* Mondeca (www.mondeca.com) has developed systems holding the equivalent
of several hundred thousand records.
* A Topic Maps system is used at Oak Ridge National Laboratory; it's part of
a system to assist officers in applying security classification rules to
documents,
which can be extremely complicated. The topic map holds the rules
and documents, and an inference engine applies the rules to the data in
the topic map.
* The U.S. Internal Revenue Service released a CD containing all of the
year's tax regulations and forms. It was indexed with a topic map.
* Starbase Corporation, now part of Borland International, created a system
that integrates several separate databases by adding a software layer that
makes them look like topic maps (Seidl 2001). The virtual topic map is
available over the company's network. This application of Topic Maps
starts to approach the vision of linking and navigating a range of different
resources distributed over a network.
* LexisNexis, which maintains an enormous online reference collection for
subscribers, has developed a prototype topic map system that makes it easier
for a person to locate information of interest (Freese 2003). Note that
this system doesn't store the main data in the topic map. Instead, the topic
map contains meta data that can be used to identify and find information
in the main data stores.
* The U.S. Social Security Administration has developed a content-management
system for policy-related documents that's accessed using an enterprisewide
topic map (Degler and Battle 2003), using earlier work that collected
references to a vast amount of Medicare information for legal users. The
original impetus was a need to handle complex medical appeals cases
more quickly. The system allows document owners to index documents
using common topics related to agency processes and terminology. Policy
subject experts add and maintain all the keywords in the topic map without
having to rely on specialist technical staff. This helps to keep the topics
familiar to people seeking content, and improves end users' ability to find
complex, interrelated content.
Here is some more info about companies and orgs working with Topic Maps.
Extensive case studies of successful Topic Map applications (IRS docs among
those):
http://www.techquila.com/bcase.html
Some company sites where they develop with TM:
* empolis: http://www.empolis.co.uk/
* Infoloom: http://www.infoloom.com/
* Mondeca: http://www.mondeca.com
* Ontopia: http://www.ontopia.net/
Good online showcase of TM, CIA Factbook as TM and the rest of it
http://www.ontopia.net/omnigator/models/index.jsp
News and info on the current state of affairs in the TM world
http://www.topicmap.com/
Small case study of using TM with RDBMS
http://www.xml.com/pub/a/2003/03/05/tmrdb.html
In fact TM is roughly equivalent to RDF and both of those technologies deal
with Knowledge Management and ontologies
(http://www.jfsowa.com/ontology/index.htm - very good book on the ontology
theory) (OWL builds on top of RDF). They provide flexible means to define
your domain model in a standartized way and make it real well adjustable to
changes in requirements, etc. Also they perfectly suitable for application
of logical inference systems. So TM is much more than just an indexing
mechanizm, people do the entire sites, content management and information
storage apps, etc. Well, everybody knows of RDF, so TM works on the same
grounds, just easier to use.
Topic maps are fundamentally about linking concepts to information about
those concepts, with the goals of promoting collocation and navigation of
the
information-finding the right information and finding all related
information
nearby. Topic maps are based on the idea of concepts-anything that can be
discussed
or thought about. A topic is a computer representation of a concept.
Concepts
are significant because of their relationships to other concepts. To model
this, topics can be related to other topics by means of associations. They
can also
contain or point to other relevant information, using the occurrence
structure
for this purpose.
Subject identity is one of the key ideas of topic maps. The subject of a
topic
map means the concept it refers to. The identity can be established in
several
ways. One is by specifying a publicly published identifier (a PSI), which
contains
a human-readable description of the subject, and possibly machine-readable
meta data as well. Another way is to point to a resource, such as a web
page, and
say that the resource "indicates" the subject. A third way is to state that
a given
resource is the subject in question. In all cases, human judgment may be
needed
to determine exactly what the subject of a given topic is. These
capabilities make
topic maps unique in their ability to specify the nature of their subjects.
Although they were originally designed to act as indexes into bodies of
information,
topic maps can store and organize data of all kinds. Their origin as
indexing systems is reflected in a tendency for topic map developers to use
topic
maps as overlays to existing information rather than as primary databases.
Well, that was a kind of resume on the subject, some interesting and useful
stuff to know and think about.
Andriy.
----- Original Message -----
From: "Andrus Adamchik" <andru..bjectstyle.org>
To: "Cayenne Devel" <cayenne-deve..bjectstyle.org>
Sent: Wednesday, September 01, 2004 1:59 AM
Subject: Re: Topic Maps: Backend Plugin for TM4J
> Hi Andriy,
>
> That's some pretty cool stuff. With both Microsoft and Apple making
> document search functionality the core of their upcoming OS releases
> (Longhorn and Tiger), and at least Apple focusing on document metadata
> indexing, I can see how such technologies can become big in the coming
> years. I wonder how many people/companies already use Topic Maps and
> TM4J? And what exactly this is used for other than book index creation?
> As unlike other search technologies, topic map creation seems like a
> very involved non-automated process... Also there is lots of developers
> on the project, but not too many downloads...
>
> Andrus
This archive was generated by hypermail 2.0.0 : Wed Sep 01 2004 - 10:58:26 EDT