RDF and Topic Maps cover a lot of the same territory. This essay exposes a first attempt at building a Topic Map for the contents of this site. It models only a few topics and associations and it probably models them badly.

The XML Topic Maps (XTM) 1.0 specification says that topic maps “convey knowledge about resources through a superimposed layer, or map, of the resources.” Like RDF, it’s an attempt to model information.

I am by no means competent to explain in any detail how Topic Maps and RDF are the same or different, but they definitely are. The same or different, I mean. Robin Cover has a page in his tremendously valuable Cover Pages archive that presents Steven Newcomb’s summary of Michael Sperberg-McQueen’s closing keynote at Extreme Markup Languages 2001. It’s about RDF and Topic Maps and it looks like as good a place as any to start. Steve Pepper’s Ten Theses on Topic Maps and RDF looks good too.

Anyway, every time I go to Extreme, a bunch of really smart people say all sorts of interesting things about Topic Maps and give cool demonstrations of how they use Topic Maps to do wonderful things. And I say, “this year, I’m going to learn about Topic Maps.”

Several years have elapsed, so I’m about on schedule. I forget what crossed my desk yesterday pertaining to Topic Maps, but whatever it was, I finally went and printed The TAO of Topic Maps and took it to bed last night for a little light reading.

This morning I hacked together a stylesheet that builds a Topic Map from the RDF repository that contains all the data about this site.

I’m starting small. My topic map has only five topic types (essays, images, people, subjects, and WordNet words) and five associations (creator, depicts, images, mentions, and subjects).

The idea is to capture the following information in the map:

There’s a lot more information on this site that could be in the topic map, but it seemed like a good idea to start small. I’ve probably managed to botch it already.

The resulting XML Topic Map, nwn.xtm, is yours to play with. All I’ve done so far is look at it in Ontopia’s Omnigator.

[Photo]

Omnigator View of an Image


There are some “unnamed” topics. That’s caused by inconsistent metadata associated with images and the fact that depictions of things identified by entries in my Palm (rather than WordNet) aren’t sufficiently annotated. I’ll fix those as soon as possible. I’ve also got some labelling problems in the map caused, presumably, by simple transposition errors or some failure on my part to understand how association labels work.

I wonder how many other things I’ve done wrong (like, shouldn’t I have more occurrences)? I also wonder if there’s an Omnigator tool for RDF or a cwm for Topic Maps.

Comments:

> shouldn’t I have more occurrences?

No, why? Occurrences are just that; pointers to an occurrence of a topic. Quite often people mistake these for properties, and lose quite a lot of semantic metadata in the process. Faceted data values makes best in associations.

I didn't stupdy your map in detail, but you've got it far better than most. Don't be too hard on yourself; it's good stuff.

Posted by Alexander Johannesen on 08 Sep 2004 @ 11:27pm UTC #

Nice! I agree with Alex that you don't really need occurrences the way you've done this, but maybe the little blurb you have on each essay could become an internal occurrence on the essay? Some of the metadata about the images (e.g. date) might have gotten the same treatment. Also, you've got a few duplicate people, which it would be nice to have merged.

Generally it looks good to me, though there's one thing I might want to change: you use subject indicator references also when the topic represents an information resource. In XTM a resource reference would be better for that.

> I also wonder if there’s an Omnigator tool for RDF

There is, and it's called Omnigator. :) Omnigator can read RDF, and it's quite easy to make it do so if you turn on the RDF2TM plug-in (go to Manage, then Plugins).

> or a cwm for Topic Maps

Not really. It would be nice if such a thing had existed, but as far as I know it doesn't.

Posted by Lars Marius Garshol on 09 Sep 2004 @ 12:14pm UTC #

A good dead-tree resource is the book, XML Topic Maps edited by Jack Park and Sam Hunting (copyright © 2003 Pearson Education, Inc.). It is basically a collection of chapters - by various authors - explaining XML Topic Maps and how they were used in various settings. The book is like the Charles Goldfarb's "XML Handbook" to the XTM world. Get it from your local library. Here's the Library of Contress Cataloging-in-Publishing Data from page iv:

Park, Jack.
XML topic maps: creating and using topic maps for the Web / Jack Park and Sam Hunting.
p. cm.
Inclues bibliographical references and index.
ISBN 0-201-74960-2 (paperback)
1. XML (Document markup language) 2. Metadata I. Hunting, Sam. II. Title.
QA76.76.H94.P376 2002
005.7'2-dc21
2002003679

Hope that's useful!

Posted by James F. Cerra on 11 Sep 2004 @ 04:38am UTC #

Just for completeness reasons: The Topic Maps Handbook (an empolis White Paper) is a further reading about Topic Maps. It is positioned between Peppers' TAO paper and Parks' book. You can download it for free at http://www.empolis.com/downloads/empolis_TopicMaps_Whitepaper20030206.pdf

Posted by H. Holger Rath, empolis on 26 Oct 2004 @ 01:45pm UTC #
Add a comment or subscribe to (existing and future) comments on this essay.
Name:
Email*:
 *Please provide your real email address; it will not be displayed as part of the comment.
Homepage:
Comment**:
 **The following markup may be used in the body of the comment: a, abbr, b, br, code, em, i, p, pre, strong, and var. You can also use character entities. Any other markup will be discarded, including all attributes (except href on a). Your tag soup will be sanitized...