[sc34wg3] Analysis of TMRM Use Cases

Kal Ahmed sc34wg3@isotopicmaps.org
Tue, 13 Apr 2004 09:42:35 +0100

Jan Algermissen wrote:

>Kal Ahmed wrote:
>>I think the biggest problem with multiple models will be uncertainty.
>It is certainly a problem of the RM draft document's prose that it seems
>to constantly fail to 'get its core point accross',...but then...noone
>has ever asked a question about it either, so it seems pretty well
>understood, or?
>Anyhow, what makes you think that the RM is about having multiple models?
>Where did you get that from?
>The RM is fundamentally about disclosing semantics (or ontologies if you
>want). It is about how to standardize the act of disclosing decisions like:
>"By an occurrence I mean:...., by latitude and longitude properties I mean:...
>I consider the combination of latitue and longitude to be a property that
>allows me to check whether two topics surrogate the same subject, etc.)
>The RM is about how to write 'disclosures of semantics', which are actually
>very similar to RDBMS schemas or RDF schemas. The major difference is that
>they are intended to work globally and that they are of much richer semantic
>To put this in other words: if I send someone else my topic map and also
>the disclosure(s) that define the semantics used in it I will be sure that
>*any* topic map engine will process the map with the exact same result.
>That's not attractive for users of Topic Maps????
Your entire explanation presupposes that Topic Maps == RM. I don't buy 
that - I never, even with a close reading of  ISO 13250 or XTM 1.0 saw 
*anything* remotely like the RM in it. So as far as I am concerned RM != 
Topic Maps.

What you propose may be attractive for some set of users, but you will 
also confuse all the existing topic maps users who, like me, did not see 
anything of what the RM described in the standard. I don't buy thea 
argument that thats just our own stupid fault and that we should have 
realised the "intent" - some of us read the standard carefully enough to 
implement it, others read it carefully enough to understand the features 
of topic maps and to explain them to others, to educate and promote a 
standard. To change it so fundamentally now will shake my faith in the 
standard - and I'm sure I won't be alone in that.

So to reiterate, I am not against the RM, but calling the RM "Topic 
Maps" seems like a leap and a big mistake.

>>Why do none of the existing implementations support this new kind of
>For the record: I have implemented the RM and I also use it with good
>success (and the joy of being able to write down the semantics I use[1]
>in a defined way) in client projects.
You are right, I apologise for leaving that out. So the question should 
have been "Why does only one implementation out of a good half-dozen or 
more support this new kind of model ?" - the problem still remains. Its 
good to see that you have done the work to prove that the RM is 
implementable. But why does only one topic map processor do all this 
stuff ? Was it really the case that everyone else implementing topic 
maps got hold of the wrong end of the stick? Or is it that neither ISO 
13250 in both published editions nor XTM 1.0 describe the RM ? If  its 
the latter, then how can WG3 really expect ISO 13250 3rd edition to be 
taken seriously when it makes such a major change in the definition of 
topic maps ?