[topicmaps-comment] back to the lists

["Bandholtz, Thomas" <TBandholtz@koeln.sema.slb.com>]

Title: back to the lists

"Dear Oasis TM community"

as a silent-for-months member of TC pubsubj and TC geolang, I feel that I owe you a statement about my (dis-) appearance.

Shortly after XML Europe last May, I had to focus (to earn my living) on finishing an R&D project "Semantic Network Services (SNS)" and on some company-internal work on KM and metadata. I followed the TM discussion, at least kept track of the topics, and sometimes I was making up for a posting, but - I still was somehow undecided ("mixed feelings", 2002) about some basic issues, and I did not find a clear position fast enough.

Now, SNS has been completed (at least more or less), and I start to look over the surface again (...is this German lingo? ...)

Last week I attended the "Open Forum 2003 on Metadata Registries" (http://metadata-stds.org/OpenForum2003/) in Santa Fe, arranged by ISO/IEC JTC1 SC32 WG2 (http://metadata-stds.org/). I followed and presented-in the "Terminology and Ontology" track, and this gave me the final inspiration to re-enter the discussion.

In (relatively) short:
++++++++++++ 1. Generally, discussion of TM should be more in the context of what's happing outside in the world of metadata, ontology, etc. As I have been confirmed once more in Santa Fe, there are many contiguous applications for TM fitting into open ends of related standards like ISO 11179, ebXML, UDDI, and Semantic Web - but people don't really know about us - though most of them are curious about our possible contribution to the game.

On the other hand, there are several approaches out there that might be adopted for open issues of our own world, like OWL, or the Extensible Resource Identifier (XRI).

++++++++++++ 2. PSI
Every metadata initiative in the world is highly interested in a well defined terminology for values, which PSI can provide. The most important thing is that there really need to be unique and persistent URI for any of the controlled terms. I think it is of less importance, to what these URI really point. All the variants mentioned in http://www.oasis-open.org/committees/tm-pubsubj/docs/recommendations/general.htm must be supported, as they are real.

The most important pre-requisite for PSI to go live are: organizations that maintain them, in whatsoever form. ISBN work sufficiently (when did you order a book last time?), as there is a strong and sustainable organization behind them.

It has been clearly agreed that we will not maintain PSI-sets ourselves as a service. So, what is our contribution?
To my opinion, machine-readable PSI will play the strategic role in the future (no problem for a machine to convert e.g. XTM into human-readable HTML).

There are (at least) two related approaches that we should include into our considerations, if we don't want to finally find ourselves in a maverick position ("small boats" are not necessarily faster, Robert :-):

a) The handling of taxonomies in ebXML and UDDI (google for "taxonomy" and one of the organization acronyms). We must fit into that and find our "added value".

(I am sure there is ... the concept of Topic Map is a valid template in the vague world of Ontology ...).
b) OASIS Extensible Resource Identifier (XRI) Technical Committee. http://xml.coverpages.org/ni2003-01-08-a.html - why aren't we in this boat? XRI is closely related to PSI, and we should get into this discussion.

Karl Best had a very generic talk at the Open Forum (not yet on-line today). He did not mention TM at all. But he was presenting a new cross-organization standards registry initiative (kind of extended xml.org - his slides will be on-line soon ...), and there was a slide showing something that looked like a topic map. Why didn't he propose it to be one? Diplomacy? Or doesn't he take us as a serious option? (see the related OASIS RFP at http://www.oasis-open.org/documents/registry_rfp.pdf).

++++++++++++ 3. What interests me most, currently, is OWL (http://www.w3.org/TR/2002/WD-owl-guide-20021104/). In my project work over the last two years I found serious deficits in the XMLization of the topic map concept.

One simpler thing is that there is no normative XML Schema for topic maps. If you want to use TM in a context like web services, any of these today require an XML Schema to be integrated. It's easy to load the XTM DTD into XMLSpy (or something) and convert it - needs some human cosmetics to get it to work, but not too much - but that's not normative! There have been lots of discussions about a TM Schema that I don't want to rehash - would be enough to make it semantically as equivalent to the DTD as possible -agreed, and "official".

What's more crucial is the lack of a serialized formal definition of the typology and constraints of a given topic map instance, such as which topic types may play which role in which association in this domain (natural lingo: "A string quartet consists of exactly 4 bowed string musicians: typically 1st and 2nd violin, viola, and cello"). Something like this was announced (http://www.y12.doe.gov/sgml/sc34/document/0323.htm) to be addressed in "ISO 19756: Topic Maps Constraint Language (TMCL)". Public discussion of further work has finally started one week ago at http://groups.yahoo.com/group/tmcl-wg/. Great!

Reading the Web Ontology Language (OWL) development  at http://www.w3.org/2001/sw/WebOnt/, I highly suspect that TM schema definitions can be completely written in OWL by defining classes like <tm:topic> as a subclass of <owl:thing>, and so provide the TM concept as an implemented OWL "library" schema that can be incorporated into any work of the Web Ontology movement. (May be integrate OWL-Wine and TM-Beer ... who'll be doing the non-alcoholics ?).

Regards,