[sc34wg3] Re: SAM: Topic Identity

sc34wg3@isotopicmaps.org sc34wg3@isotopicmaps.org
Thu, 03 Jul 2003 13:33:48 GMT

Murray Altheim writes:

> Lars Marius Garshol wrote:
> > * kal@techquila.com
> > |
> > | Rereading the SAM with CXTM glasses, I notice that the Topic
> > | information item does not require any form of identity. Each of the
> > | properties [source locators], [subject indicators] and [subject
> > | addresses] may be empty as can the [names] property.
> > 
> > I've been uneasy about this myself, and I do feel that it might be
> > that a topic should be required to have some form of identity. RDF
> > does not require this, but even so it would make life much easier for
> > upper layers (TMQL, say, as well as any other technology that uses
> > topic references) if this were required.
> Actually, while I agree in principle with the desire for a stated
> identity for all topics, by virtue of having a <topic> element in
> an XTM document, the topic exists and has identity. This would be
> the equivalent of the LTM and XTM expressions
>    [dog]
>    <topic id="dog"/>

In both cases "dog" becomes the fragment identifier in the [address] of a
[source addresses] property value.

> I think this is quite all right, actually, especially as for authoring
> purposes, one may *need* to be able to create a valid but empty topic
> in order to later populate it. And for simple demonstrations, where the
> topic's ID is considered as its identifier, there isn't a huge problem
> to be solved here (IMO), e.g.,
>    chases([dog] : pursuer, [cat] pursued])
> Or are you guys talking about something else and I'm just missing it?

My issue was that SAM would allow

Which the syntaxes do not allow. Of course, one could create a syntax which
did allow this but unless there is a convincing reason why such constructs
should be supported by the model, the model should be tightened up to
prevent it.