Announce: Topic Map Standard out for Final Committee Draft Ballot

W. Eliot Kimber eliot at dns.isogen.com
Sun Nov 15 14:51:09 GMT 1998


At 08:48 AM 11/14/98 -0800, Dave Winer wrote:
>What kind of applications would we use the Topic Map structures for? 
>
>It always helps me to understand this kind of stuff if I can understand a
>compelling application for it.

A topic map consists, fundamentally, of two kinds of things: topics and
associations. A topic is an object that represents a single rhetorical
topic or subject. For example, "XML parser" might be one topic in a topic
map about XML. A topic serves to associate the abstract idea of the topic
with occurrences of that topic:

<topic xml:link="extended" role="topic" id="xml.parser">
<name>
<basename>XML Parser</basename>
</name>
<occur xml:link="locator" role="parser-instance"
       href="http://www.jclark.com/xp"/>
<occur xml:link="locator" role="parser-instance"
       href="http://www.microsoft.com/msxml/">
</topic>

Notice that this serves to impose the semantic label "XML Parser" onto the
occurrences addressed by the occur element.  Thus a topic can assert that a
given object is an occurrence of some kind of thing.  This lets you
construct a classfication or descriptive layer on top of existing data.
The topics essentially represent opinions about the data.  Different topic
map authors might express different opinions about the same data. Because
the form the opinions are expressed in is standardized and consistent
(topics), they can be reasonably compared to some degree.

Because the topics are expressed formally as hyperlinks (here using Xlink,
but also doable using HyTime), they are naturally navigable using whatever
hyperlinking support you have lying about (e.g., HyBrick, PHyLIS, etc.).

Associations relate topics to each other. To continue the XML topic map
idea,  might have a relation "standard-interface-for" that I use to relate
the topic "XML parser" to the topic "SAX":

<assoc role="standard-interface-for" xml:link="extended">
<assocrl role="parser" href="#id(xml.parser)" xml:link="locator"/>
<assocrl role="interface" href="#id(sax)" xml:link="locator"/>
</assoc>

In many ways, this is like RDF: you can impose properties onto data objects
and relate data objects together using typed links.  It may be that topic
maps are one way to express RDF abstractions, I don't know (I don't know
enough about RDF).  But while RDF seems to be designed primarily to support
the addition or representation of metadata about objects, topic maps are
designed for the creation of knowledge bases imposed on data of any type,
and, in particular.  In any case, Topic Maps are not intended to compete
with RDF--they are, in essence, different views of the same abstraction:
objects with properties and relations among them.

So what would you use topic maps for? I think one compelling use is as an
annotative or descriptive layer over things like encyclopedias,
dictionaries, databases, and the like. They might be used to enhance
management information systems by providing a simple but rich and
standardized way to capture analysis applied to existing data, such as
market reports, sales numbers, etc.

Topic maps can be a way to augment search and retrieval by providing a form
of index over a larger, more amorphous body of data. 

Many documents can be turned into topic maps simply by labeling the
existing components as topics. For example, you can think of a command
reference document as a topic map where every command description is a topic.

The reason for standardizing this concept is that it lets you build generic
topic map engines that understand the specific properties of topics and
associations and can therefore manage knowledge of those properties in a
crisp and efficient way, making the information available to processing
systems. It also allows the meaningful and automatic merging of topic maps
because it's clear how the components of each relate to each other as objects.

I'm not sure I've answered the question very well, but maybe I've provided
enough of a taste for what topic maps do that applications will suggest
themselves.  I've left out a number of important and interesting details in
the discussion above, but I think I've conveyed the flavor.

Cheers,

Eliot
--
<Address HyTime=bibloc>
W. Eliot Kimber, Senior Consulting SGML Engineer
ISOGEN International Corp.
2200 N. Lamar St., Suite 230, Dallas, TX 75202.  214.953.0004
www.isogen.com
</Address>

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:majordomo at ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)




More information about the Xml-dev mailing list