[topicmapmail] multidirectional property of associationrole

[Lars Marius Garshol]

* Lars Marius Garshol
|
| I have never seen your software, beyond a single low-resolution
| screenshot, so how could there fail to be "mystery"?

* Bernard Vatant
| 
| I'm well aware of that. We don't have any "light download" version
| of the software, since it requires Oracle + specific installation.
| That was a deliberate original choice, on which Jean could expand
| more - I plead not guilty on that :))

There's no guilt involved. It's just that you made it sound as if I
should have known better, when in fact I did not have occasion to know
anything at all.
 
* Lars Marius Garshol
|
| Both Ontopia and empolis use a core object model, based on the
| implicit data model of XTM.  These models differ slightly, but are
| in essence the same.
 
* Bernard Vatant
|
| That's where we differ, and I am not sure you caught it. The
| "implicit data model of XTM" is not the core object model of
| Mondeca, and Mondeca never pretented it was.

I know, I just saw red when I heard the same old syntax nonsense yet
again. 
 
| Mondeca has supported Topic Maps specification and XTM development
| and still does inside OASIS TCs, but never pretended that its core
| model was fully conformant to anything apart its own original
| conception of what a "semantic graph" is - which happened to fit
| nicely with, say about 90%, the topic map "implicit model" (whatever
| that means)

This is an honest answer. To me this is perfectly OK. Any vendor
and/or developer must choose what to conform to and what not to
conform to, and as long as they are honest about it there is nothing
to blame them for. (Though one may choose not to buy their software. :)

You can hardly blame people for being uncertain about exactly what it
is Mondeca does, however, when the web site keeps talking about how
what you have is topic map technology, and yet when people look closer
at it it looks rather different.

It's good that you've now been open about it.
 
| Let me be more precise, so that you really catch my point. 

Thank you. I obstinately refuse to catch it as long as you are, ahem,
being imprecise. :-)

| Mondeca has always considered *topic maps* like one form of
| structures they could/would manage. That's not only a question of
| *format*, but really a question of *model*. We know pretty well we
| have internal features that do not conform to the TM model. For
| example we can "filter" (scope) and name every single object, be
| they nodes or edges, and that goes down to allowing scoping types
| and naming associations, and some more of the same "heretic" kind.

Well, after the advent of reification this does not sound like heresy,
except perhaps scope on topics.

| If we want to export XTM conformant files, we have to be careful not
| to use some features of the software.  That's all. To go to the core
| of the debate, our graph searchers have always considered the topic
| map model as unnecessarily convoluted and constrained in many
| aspects. And basically I sort of agree with them :))

In that case you should come out in the open with your arguments. We
are defining the model now, and it would be better to hear your
arguments before rather than after it is all cast in stone.
 
| There again I'm not seeking to argument, just to explain. We had a
| strong core model from the beginning, that we still consider to be
| more generic than the XTM data model, and even the topic maps
| implicit model. It is a mathematical model standing on graph theory,
| which is IMO a quite reliable support. So why would we change, and
| add constraints to conform to a model that would restreint our
| capacity to deal with others?

That's up to you.
 
* Lars Marius Garshol
|
| It is *crucial* to understand that topic maps are not about
| syntax, they are a *model* and if you want to implement topic maps
| you must follow the model, regardless of whatever syntaxes you
| happen to support or not support.
 
* Bernard Vatant
|
| *You* want to *implement* topic maps. We just want to be able to
| import/export them.

As long as you say so and are clear about it I have no problem with
that. 
 
* Lars Marius Garshol
|
| What are you going to do when TMQL and TMCL appear, based on the
| Standard Application Model, and assuming that all implementations
| follow it? What will you do if Holger's idea of TMSL-T becomes
| reality, since that, too, will have to build on the SAM?
 
* Bernard Vatant
|
| We're not yet there.

That's fair enough. Do you intend to ever get there, or are you
happier where you are?
 
| Hmm ... I think we have collapsed two problems
| 
| The first one is labeling in some standard way roles and
| associations for internal TM management and exchange. It is for that
| one that I think we should stick to substantive non-directed labels.

You are right that we were discussing both these issues. Frankly, I am
not really interested in this first issue. Internal names are
internal, and what you call your associations as you are modelling the
topic map is your business.

(Substantive is something else, BTW. :-)

| The second one is the rendition for end users, which, I agree with
| you, should be the most natural possible.

Good. Have you looked a the Omnigator? I think it's probably necessary
to do so to fully understand what I mean. Look at the associations
from both sides.

<URL: http://www.ontopia.net/omnigator/models/topic_complete.jsp?tm=opera.xtm&id=puccini >
 
| What is natural and non-natural anyway? When you read a table with
| column titles "person" and "birthplace", that is not maybe *natural*
| language but it is crystal clear for even the most basic information
| user, from age 7 and up ...

That's a good point. We don't display our associations that way, so we
chose a different way to select the names. If you use tables it means
that whatever solution we choose does not apply to you.

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
ISO SC34/WG3, OASIS GeoLang TC        <URL: http://www.garshol.priv.no >