Antone Roundy wrote:
> 1) If I use the URI "http://www.geckotribe.com/atom"; to talk about
> the Atom format, and you use "http://bobwyman.pubsub.com/atom"; for
> the same thing, how is a client to discover that these are the
> same subject?
Do you really mean the same "subject" or did you mean the same
"topic" or "category"?
If you meant the same "topic/category, then XTM would encode this
something like the following: (Note: I'm not an XTM guru. Please forgive
errors in syntax...) The intent of the topic definition below is to say that
both subjectIndicators refer to the same topic "The Atom Format." The
<topic id="atom_format">
<baseName><baseNameString>The Atom Format</baseNameString></baseName>
<subjectIdentity>
<subjectIndicatorRef xlink:href="http://www.geckotribe.com/atom"/>
<subjectIndicatorRef xlink:href="http://bobwyman.pubsub.com/atom"/>
</subjectIdentity>
</topic>
Given that you might be supporting people working in multiple
languages, you might find it handy to specify the name of the mapping from
the subjectIndicators to topics in more than one language. Perhaps, both
French and English... If that was your desire, then in XTM, you would
specify language by establishing a "scope" for the baseName:
<topic id="atom_format">
<baseName><scope><topicRef xlink:href="#en"/></scope>
<baseNameString>The Atom Format</baseNameString>
</baseName>
<baseName><scope><topicRef xlink:href="#fr"/></scope>
<baseNameString>Le Format Atom</baseNameString>
</baseName>
<subjectIdentity>
<subjectIndicatorRef xlink:href="http://www.geckotribe.com/atom"/>
<subjectIndicatorRef xlink:href="http://bobwyman.pubsub.com/atom"/>
</subjectIdentity>
</topic>
The definition above could, of course have been done in two
declarations -- even in different files. However, the two would still be
linked by the common subjectIndicators.
> 2) What's to prevent two people from using the same URI as the
> subjectIndicator for different subjects? For example:
> "http://www.xerox.com/"; -- is that the company Xerox, or copy
> machines in general?
Nothing other than a desire to make interop work. My assumption is
that rules would say that people should create globally unique URI's for new
subjectIndicators (tagURL's would work fine). Also, we should probably have
a registry for these things. We'd be happy to fund and manage such a
registry if subjectIndicators are accepted in Atom.
In any case, I think we'll eventually find that people would tend to
use commonly accepted subjectIndicators and then map them to "local" topics
or categories. Getting conformance on subjectIndicators is much less "bad"
then getting conformance on classification schemes since there is less
"subjective" content to a subjectIndicator then there is to a
classification. Accepting common classifications or topics can often force
one particular point-of-view on others -- not good.
bob wyman
