[sc34wg3] Topic Maps -- Reference Model

Patrick Durusau sc34wg3@isotopicmaps.org
Wed, 08 Jun 2005 19:52:36 -0400


I sent this post while in Amsterdam, 26 May. Apologies to Ann for the 
late delivery of this response!


Anne Marie Cregan wrote:

> Dear WG3 Topic Mappers,
> Firstly let me say what a delight it has been to meet a fair 
> proportion of you in Amsterdam and to get to know you and learn more 
> about the Topic Map work.  I am very appreciative to those of you who 
> listened to my presentation on OWL and Topic Maps on Monday, and who 
> gave their opinions and feedback.  Some of the issues will no doubt 
> need to be considered on a more fundamental level, which is where the 
> TMRM comes in and might provide some guidance.
We were delighted to have you at the meeting and look forward to your 
future attendance and contributions!

> However, being new to the Topic Map WG3 community, I am somewhat 
> unclear about what the TMRM is all about, and attending the TMRM WG3 
> meeting on Tuesday left many of my questions unanswered.  It seems to 
> be potentially a very interesting piece of work that I might like to 
> contribute to, but I'm not sure where to start at the moment.
> So, I would like to invite anyone so inclined to respond to the 
> following questions and hopefully shed some light on the matter.  
> Perhaps t his may generate some discussion which woul be helpful for 
> everyone, as it may help to clarify the various positions that are out 
> there, and perhaps to identify where the common ground is.
> The questions I would invite you to respond to are:
> 1. What do you see as the purpose of the TMRM?

The TMRM seeks to provide a non-data model specific formalism that can 
be used to disclose different ways of indicating subjects.

> 2. What form do you envisage the TMRM taking? I know there is a 
> mathematical model that Robert Barta is working on, but do you see 
> there being other parts, and if so, what form would they take, what 
> would the function of each part be, and how would they fit together?
The TMRM has a fairly long history which I will not repeat or summarize 
here. Suffice it to say that various participants in the topic maps 
community have held a range of views as to its relevance to topic maps 
and its utility. Those views range from thinking that it is the point of 
the paradigm to it being irrelevant. I am hardly disinterested in the 
issue so read the following with the understanding that I think it has a 
great deal of relevance and makes the paradigm more powerful than it 
would be without it.

The latest terminology discussed as a working term for an expression of 
the TMRM is "subject map disclosure." The prior term for that was Topic 
Maps Application. Somewhat unhelpfully, the latest draft defines that as:

   A set of constraints, disclosed in conformance with the requirements
   of this International Standard, on the
   structure, processing, and interpretation of the properties of
   Subject Proxies.

The reason why the draft does not report a "way" to do that is that any 
choice of a means of expressing "constraints" of necessity involves 
choices and tradeoffs that would exclude some ways of indicating 
subjects. That is to say that one could have a "subject map disclosure" 
(to use the new term) written in any syntax or method that was 
appropriate to a particular system.

What makes the notion of "subject map disclosure" powerful can be seen 
if you thiink about Steve Pepper's excellent demonstration of RDF -> 
topic map earlier this week. All well and good to do it, but how does 
anyone else make the same transformation? In other words, where is the 
mapping that other topic map software could follow to reach the same 
result? The RDF data model, and the manner in which it indicates 
subjects, is different from that of the TMDM. In order to have a 
meaningful integration of such a mapping with other mappings of RDF, it 
is necessary to know how each of the mappings were performed.

While RDF is a subject based technology (I am eliding over differences 
in terms of what that may or may not mean) and is perhaps a little more 
transparent to map to the TMDM, other approaches to indicating subjects 
are less transparent. Consider how to map the subjects found in a COBOL 
file for example. (75% of all business data is still held in COBOL 
according to a recent Gartner report, makes one wonder about claims that 
XML is sweeping the world.) The mapping of the subjects and what data 
indicates what subject is far from apparent looking at a COBOL file.

> 3. What do you  think is the best way to "ground" subjects? That is, 
> to unambiguously and unequivocally identify them in a a way t hat is 
> useful and convenient for the rest of the TM-related methodologies?

By way of explanation for other readers, Anne and I had a discussion 
about "gound[ing]" and "to unambiguously and unequivocally identify" 
subjects  earlier today. In TMRM terms, there are many ways that are 
thought to "ground" or to "identify" subjects and the TMRM makes no 
choices in that regard.
What the TMRM has been attempting to do is state what the "disclosure" 
requirements are for saying the answer to that questions. Note it has 
not been trying to "say" any answers, merely to say what has to be 
disclosed for others to have a useful answer to the question. There are 
many technology standards that tell users what it thinks should be said. 
The TMRM is one of the few (perhaps only one) that asks for disclosure 
of how you are going to say what you want to say.

> 4. What would you ultimately want to be able to do with the TMRM? That 
> is, do you have any use cases in mind that you envisage the TMRM would 
> address?
The question of what the TMRM "does" or "does not" do is a fairly 
complex one. I have a short paper that is being edited on that question 
and will try to post it in the near future.

The reason I mention that issue is that I think the better question is 
what does the TMRM enable?

One thing it would enable would be a mapping of RDF, a disclosure of the 
mapping of RDF to TMDM (which assumes a mapping of TMDM as well).

Another thing it would enable would be the mapping of other means of 
indicating subjects, either to the TMDM or between means of indicating 

Another example would be capturing event information in environments 
where auditing is a necessity, such as in secure computing environments 
as well as in medical environments.

> I appreciate that there are probably no definitive answers to these 
> questions right now as the model is still at an early stage of 
> development, but I am interested to learn everyone's viewpoint and 
> perhaps through this process, to clarify where the areas of consensus 
> and of disconnection are.  Also, there may be more questions that need 
> to be asked that are not on the list above, so please raise them if 
> you feel this is the case.
Appreciate your interest and taking the time to ask the questions. Hope 
the answers are of some help.

Hope you are having a great day!


Patrick Durusau
Chair, V1 - Text Processing: Office and Publishing Systems Interface
Co-Editor, ISO 13250, Topic Maps -- Reference Model
Member, Text Encoding Initiative Board of Directors, 2003-2005

Topic Maps: Human, not artificial, intelligence at work!