[tmql-wg] Re: TMQL Problem (scoping occurrences).

[Michael Chapman]

On Monday 08 November 2004 4:42 am, Lars Marius Garshol wrote:
> Hi Michael,
>
> * Michael Chapman
>
> I think this is really more of a question for TOPICMAPMAIL than for
> tmql-wg, where we really want to discuss TMQL more than other things.
> At the moment it doesn't seem that there will be any direct connection
> between LTM and TMQL, BTW.

Apologies, but having started it I had perhaps better end it here.

> Michael, I have to admit that I'm a bit confused. What is the problem?

> Your second LTM example isn't valid LTM (or even topic maps), but I
> think what you really want to say there is
>
>   {rodin, short-description, [[...]]} / lang-fr
>   {rodin, short-description, [[...]]} / lang-en
>   {rodin, long-description,  [[...]]} / lang-fr
>   {rodin, long-description,  [[...]]} / lang-en

Basically I have a situation where there are many descriptions. Some of these 
have language variants. Some do not.

I do not wish to 'pollute' the occurrence_role_type by giving it a (to me 
gratuitous) serial number or other tag (as in "description-1" or 
"short-description") merely to lump the language variants together.

My conclusion from various implementation attempts with a real data set were:

1. It is possible to workaround the problem by giving serial numbers (or 
extra naming elements, e.g. "short-", "long-") to the occurrence type role.
This though, then, causes problems with handling occurrence role types as one 
type is artificially sub-divided. 

2. The scope (in this situation at least) is _not_ the scope of the 
occurrence as a whole. 
The scope is the scope of the resource data (or the locator).

Obviously either I'm totally up a blind creek . . . or I'm proposing the 
notation should be modified . . . *

Thanks for your comments,

	Michael

* To keep this brief: That Occurrence scoping should be closer to Topic 
notation than Association notation. MC.