ISO meeting in Atlanta, day 1
This is an unofficial report from the winter meeting of ISO JTC1 SC34. This is the committee responsible for SGML, DSDL (XML schema language framework), and Topic Maps, which usually has a meeting late in the year in conjunction with the XML USA conference. This year it's in Atlanta. The report is mostly about what Working Group 3 is doing, since this is the Working Group focusing on Topic Maps.
The first item to be discussed was CXTM, where no progress had been made since Kal Ahmed resigned his editorship. New editors had been appointed, but none of them really felt confident enough to carry the work forward. The solution to this was to appoint me as co-editor. I will work with prof. Jaeho Lee (of Korea) to produce a new draft soon. The main thing that needs to be done is basically to update CXTM to follow the latest TMDM updates.
Graphical Topic Maps notation (GTM)
Professor Lee presenting
Professor Lee has done work on a graphical notation for Topic Maps, which he calls GTM, and he wanted to present it to WG3 to encourage WG3 to take it up as a new work item (that is, to make a standard). This has been discussed for a long time, but always put off because there was felt to be too little previous work in the area, and because there were no volunteers to take it on. Professor Lee is trying to change this, in part by reviewing existing work, and he seems to be suggesting that he could volunteer for this. (I'll link to the slides once they are published somewhere.)
A big open issue was whether or not to make GTM support only the ontology level, or whether it should also support the instance level. The general feeling was that supporting the instance level would also be desirable.
Professor Lee's survey of existing work covered:
- VisualScript, by Steve Pepper and Graham Moore. This was done in response to a question by the vendors of SmartDraw whether Topic Maps could be modelled using SmartDraw. It was never published anywhere, but professor Lee got it by email from the creators. VisualScript only supported the ontology level, but later work also covered the instance level (Steve did not consider this part of VisualScript, though the slides don't make this clear.)
- Conceptual Modeling of Topic Maps with ORM versus UML, Are D. Gulbrandsen. This was presented at TMRA'05 about a month ago.
- A simple UML-based notation for Topic Maps, by Kal Ahmed. This was presented at an XML conference a while back.
Other examples of similar work were also brought up, like Ontopia's Vizigator, and some German work presented at the German conference in Darmstadt.
Professor Lee has a project funded by the Korean Standards Association, and so is interested in taking this work on.
It was agreed to put together the text for an NP (New work item Proposal) by Monday (day 3).
We went through the Japanese comments on TMDM, together with Patrick Durusau's comments. Most of this was straightforward, but eventually strayed into a long discussion on clause 7.5, which defines PSIs for all terms in the glossary of TMDM. There was a general feeling that these PSIs are useful, but also that defining them creates enormous potential for problems. We broke off for lunch without really completing this part.
Chairman and secretary (secretary not really sleeping, but actually leading the meeting :)
The plenary is the official opening of the ISO meeting, where all three WGs attend, and the formal, organizational aspects of the meeting are handled, such as liaison reports, etc. The major item here was the ISO standardization of the OASIS Open Document Format (ODF).
Another item that came up was an amendment to ISO 9573-11, which will be a RELAX-NG schema for ISO standard documents. This is good news, as many different editors have been using different versions of a schema originally created by James Clark for RELAX-NG. Having this standardized will be a good thing for all concerned. The editors have created a SourceForge project for this.
The rest of the day was spent in heated discussion over clause 7.5, which identified a large number of issues, and some possible solutions. There was little in the way of clear consensus at the end of the discussion, so the first item on the agenda for the following day was to continue the discussion of clause 7.5.
After the meeting, Patrick Durusau invited the entire committee to dinner consisting of traditional southern cuisine at his home in Covington, just outside of Atlanta. As far as I could tell, the entire committee was there, and the dinner was much enjoyed by all.
The second ISO meeting of 2007 was held over three days in Montréal, Canada, in conjunction with the Extreme Markup conference
Read | 2007-08-07 16:58
The last day of the Seoul meeting also had the closing plenary in the morning, with lots of excitement around who would vote what regarding the OpenXML submission from ECMA and Microsoft
Read | 2006-06-02 04:21
Are D. Gulbrandsen - 2005-12-02 13:50:25
GTM looks interesting, but my intuition is that there's no need for a TM-centric graphical notation for ontology modeling. I think modeling on the conceptual level should be independent of whether one would like to implement the ontology as Topic Maps.
Gabriel Hopmans - 2006-01-18 16:05:37
I want to react on Are's comment. GTM for ontology modelling seems to me very relevant. It would be really useful if you can use the graphical notation when discussing the ontology with a group of people and 'in the end' share it with them. The disadvantage with UML is that you spend a lot of time with it and in the end you still have to start develop/build something while implemented as topic map you already have a good start (if not more then that).
Lars Marius - 2006-01-22 16:32:02
I think the discussion aspect can be covered by both GTM and existing modelling languages (UML/ORM/ER), and the benefit to existing modelling languages is that tools for those already exist and are quite widespread.
I do agree with Gabriel, though, that the disadvantage to non-TM modelling languages is that you have to go through some kind of conversion to turn the model into a topic map. Especially if you want to keep tuning the model this can be awkward. Another potential disadvantage is that GTM can offer proper support for features specific to TMs (scope, PSIs, the name/occurrence/association distinction, etc).