Wikis As Topic Maps

From Eugene Eric Kim

Collab:Topic Maps have the potential to be a wonderful tool for organizing information. The data structure is great, but I have yet to see a good user interface for categorizing data using Collab:Topic Maps.

Wikis have the potential to be that user interface. Wikis are widely used for publishing content both collaboratively and individually, because of their simplicity as a hypertext authoring tool. More importantly, their data structure already bears close resemblance to Collab:Topic Maps, as both can be represented as graphs. Collab:WikiWords often map quite naturally to topics. Links listed on Wiki pages can be thought of as occurrences. More importantly, Forge:Backlinks to Wiki pages can also be thought of as occurrences.

Wikis are used in a number of different ways, most notably as Content Management Systems and as collaborative spaces. When they are used in conjunction with blogs, they act as Collab:Topic Map-like support structures.

I'll use my own blog as an example, but I want to point out that there are several other integrated blogs and Wikis out there. Collab:Chris Dent uses the same system as me in essentially the same way, except he uses MovableType as his blogging tool (I use blosxom). Bill Seitz has hacked ZWiki into what he calls a Wiki Log. Socialtext uses KWiki as the basis for its tool. All of these systems exhibit the same behavior described here.

When I write on my blog, I often use Collab:WikiWords for names of people and concepts. This behavior is quite natural and usually requires little thought. If those Collab:WikiWords exist as pages on my Wiki, then the words link to the appropriate page. If no page exists, I can create the page, just as I would from the Wiki itself.

Over time, I may use the same Collab:WikiWord in several different entries. If that page exists on the Wiki, and if I look at the Forge:Backlinks to that page, it will list all of the blog entries mentioning that Collab:WikiWord. In other words, it maintains a list of occurrences automatically; all I have to do is use Collab:WikiWords as I write.

If you look at the pages on this Wiki, all of the Collab:WikiWords represent concepts that are relevant to me and my writings. All of these pages emerged organically in the process of Story Telling. If you look at the Forge:Backlinks for a Wiki page, you can see all of the contexts in which I use the Collab:WikiWord. For example, if you look at the backlinks for "DougEngelbart", you can see all of the blog entries where I mention him. Those entries tell just as much about the man as the content on the Wiki page itself.

This capability works best over time, where concepts may occur repeatedly in unexpected contexts, and in a collaborative blog and Wiki environment, where it can reveal different usages and interpretations of concepts by different people, thus leading to shared collective understanding.

Using Wikis in this way allow people to create Collab:Topic Map-like structures in an organic, almost transparent manner, providing the benefits of Collab:Topic Maps without requiring people to dramatically change the way they work or the tools they use.


The Wiki data structure resembles, but is not equivalent to Collab:Topic Maps. In order to make it equivalent, Wikis would need Typed Links, which would enable people to define Collab:Topic Map associations. For example, you would be able to say that Turn Off Your Computer isA Collab:Patterns, where "isA" is a typed link between the two Wiki pages. With this feature (which we eventually plan on implementing in Forge:PurpleWiki), you could export Wiki content as a full-fledged Collab:Topic Map.

Some Wikis have disabled the Collab:WikiWord feature (most notably, Wiki Pedia and Socialtext). I understand the reasoning behind this, but I don't think it's a good idea, especially in the context of Wikis As Topic Maps. more on this later.


Collab:Jack Park's NexistWiki was the first attempt -- to my knowledge -- to generate Collab:Topic Maps using Wikis. RDFWiki does a similar thing with RDF, and it's based on the same observation -- that the Wiki's graph-like data structure resembles RDF.

In 2002, I wrote a paper, "Towards a Standard Graph Model for the Open Hyperdocument System: Enabling Interoperability Between Collaborative Knowledge Applications," that shows that most knowledge tools use a graph-like data structure, and that explores the benefits of standardizing on a metamodel for describing and manipulating graphs. Ken Holman presented this paper at the Extreme Markup conference in August 2002.

Also see Markup Free Auto Linking Wiki.


You could use PeriPeri's metadata syntax to create typed links.