What's wrong with Topic Maps?

This might seem like a strange question given the recently succesful conference Topic Maps 2008 and the general embracement of the technology in Norway. The underlying worries are tied to the latter: will TM remain a Norwegian speciality or will it get more recognition international also? This is therefore a follow-up of the panel dicussion closing the conference.

First I have to state my own relation to TM. I am by no means an expert of TM - I will caracterise myself as an advanced user. I first learned about TM in 2001 when I invited Lars Marius Garshol to talk about it at our annual IT conference IT-forum. I was immediately attracted to the idea but first in 2004 I was able to work with TM on a project base. I have found TM to be especially useful as a pedagogical tool for describing and explaining information structures.

TAO and portals
But why hasn't TM caught the same attention outside Norway and what can be done to help that? In Norway the adoption of TM has been particularly in the public sector as a framework for building portals. TM suits the portal metaphor very good and it is intuitively conceived as the "right" foundation for building semantically strong portals.

In his closing note, Steve Pepper seemed to be uncomfortable with the fact that TM is first and foremost used as a portal platform. His reason for doing so is that TM is so much more than portals, which he describes is the TAO (topics, associations and occerences) part of TM. I have also heared this "concern" been raised by other TM people. I think this is the wrong strategy: TM should be pushed as a portal framework with no concerns - portals are the killer application for Topic Maps.

Lightweight, open tools
At the conference I raised the question of lowering the barriers for using TM. Maybe the question was a bit confusing - the answers partly indicated that. So I will try to clear up a few things (around here :-). I think there is a need for easier, more accessable tools both for developers and end users. I will start with the developers.

In order to give TM a real push forward I think the TM community should join efforts to develop open source TM engine(s), modelling tools and CMS's. I know Bouvet has developed an OSS TM engine (and CMS) called ZTM, but they have been very reluctant to market it, at least after my opinion. I don't know the ZTM software so I don't know if it is the right piece of software to build on.

In addition to an OSS engine there is also a need for a simple, visually enhanced modelling tool. This is by no means an easy task but I think it would do very much to boost the interest for TM. A part of this effort could also be to provide ready-made ontologies easy to adopt for others. As a matter of fact we plan to do this with our own ontology (underlying www.vestforsk.no) in a more formal way by handing it over to Abelia, our own employers organisation. It can also be done in a very simple and straight forward method by including a link to an xtm file on the homepage.

Finally there is a need for open source CMS's with TM support. This is probably an easier task, and some initiatives have already been done. I think it would be a very good move to build TM support for the most popular OSS CMS's, like Drupal, Plone, Joomla, DotNetNuke - just to name a few.

Web 2.0 and end users
Steve Pepper mentioned adoption to the web 2.0 movement as an untapped potential for TM. I definitely think he is right, but there is a need for initiatives in order to get going. I think examples like Freebase, as I have mentioned in another blog post, can serve as a good example. This will provide users with a minimal formal ontology as a starting point onto which can be built whatever the users have in mind.

Other use cases can be TM as a blog supporting tool to have richer blogs and support findability.

Marketing
David Weinberger's point that Topic Maps are not maps and as such is vulnerable to misunderstandings is real. I have been met with questions like "how can this GIS system be used?" more than once.

David used a US company specialising in tools for faceted classification as an example. They don't market their product as faceted classification but as guided navigation. He also pointed out that the name of the standard and the marketing name of course doesn't have to be the same.

Merging
Some word on merging as well, as Steve pointed out represents a huge possibility in TM, but as of now is almost unused. It is quite a paradox having all these semantically rich TM-based public portals with none of them speaking to each other. They are indeed semantically rich islands.

The quest for a good, scalable method for distributing and using PSIs is important. Finding a clever method for handling PSIs is crucial, but I think even more important are the organisational aspects of this challenge. I think these aspects have been greatly underestimated. Who is going to be responsible for a PSI? Why should it work in TM/on the web when it doesn't even work in real life? Maybe the solution is the twist Stian Danenbarger gave to David Weinbergers famous quotation small pieces meaningfully joined. That is, the ambitions must be scaled down quite a bit.

Irrespective of the above; better working examples of useful merging of TMs are highly sought after (italian operas merged with country codes just isn't relevant enough).

Finally, to answer my initial question: Nothing's wrong with TM but it does not get the recognition it deserves!