[topicmapmail] Contextualized Topic Maps.

[Murray Altheim]

Carlo Moneti wrote:
>   * Murray Altheim wrote:
>>I think the idea of scoping topics is a bit of a non-sequitor, and
>>is perhaps a mistaken understanding of the Topic Maps paradigm.
>>In the real world as well as in Topic Maps, topics simply *exist*
>>(in TM, Topics are proxies for their subjects). I believe what
>>you really want is the ability to provide a context in which those
>>Topics play a role in an Association.
> 
> * Carlo Moneti wrote:
> 
>>Is scoping topics really unapplicable? Suppose I have a repository of 
>>documents, and I want to see the "reviewed" ones and not the 
>>"unreviewed".  What would be the recommended approach?
> 
> *Thomas Passin wrote: 
> 
>>Scope the base names.  When you "see" a topic, normally you "see" its 
>>name in a list or display.  By filtering on the scope, you can prevent 
>>its appearance, thus effectively removing it temporarily.  Similarly, you 
>>can scope occurrences.
>>
>>In your example, you could also use a "status" occurrence to track the 
>>reviewed status of the topics for the documents, and filter on the 
>>status.  That is most likely how I would do it.  I think of scopes as 
>>being less emphemeral than a status that could change.  To me, they are 
>>more about the nature of a (whatever), and less about its temporary 
>>state.  But that is not actually in any spec, it is just my view.
> 
> Exactly. This is why I asked "Is scoping topics really unapplicable?" 
> Scoping topics can work. But scoping a basename for a status seems to 
> stretch (pollute) semantics; perhaps for occurrences as well. What is the 
> recommended approach? Any thoughts? Experiences?

Carlo,

I think somebody has already mentioned unary relations. Your problem
is similar to one I have in my own application, where I'm visualizing
fairly sizable ontologies that have an underlying set of "utility"
Topics that I don't want visualized. I have a "Hidden" PSI which is
used for this. Basically, any Topics that are an instance of a Hidden
Topic (i.e., they contain a reference to the Hidden PSI in their
<instanceOf>), my visualization engine doesn't show them.

Now, as I've mentioned before, this can be done one of two ways,
the first as above, by including a reference to a PSI in <instanceOf>,
or second, by making an association between the PSI and the Topic. In
reality, they both do the same thing:  assign a characteristic to a
Topic. That's what you want to do, assign a characteristic. I agree
that assigning a name or an occurrence is probably not the most
semantically clean way to do this. Stating that a given Topic is
"hidden" or "reviewed", i.e., is an instance of the class of Topics
that are "hidden" or "reviewed" would seem to be the way IMO.

I've built a whole Faceted Classification scheme as a layer in my
TM-based ontologies, where I could assign a Topic a facet, which
would essentially be the same thing as I'm recommending above, just
that the FC model has a bit more complex (and interesting) semantics,
such as an inheritance hierarchy.

Murray

......................................................................
Murray Altheim                    http://kmi.open.ac.uk/people/murray/
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK               .

    LEO (July 23-Aug. 22): When he first invented the printing press
    in the fifteenth century, Johannes Gutenberg used it to fashion
    frivolous little things like sets of playing cards. Only later
    did he put it to work mass-producing books, turning it into a
    revolutionary tool for disseminating information. I foresee a
    similar evolution for you in the coming weeks, Leo. In the early
    going, you'll employ a wonderful new resource in a relatively
    impractical way; or maybe a big, fresh idea will first appear in
    a trivial context. Not to worry, though. After the initial false
    starts and wasteful experiments, you will ultimately discover an
    elegant use for your innovation. Be patient.      -- Rob Brezsny