[topicmapmail] Subject Identifiers metadata

[Kal Ahmed]

On Wed, 2004-05-19 at 03:12, Thomas B. Passin wrote:
> Nikita Ogievetsky wrote:
> 
> > I am also in out on this list, but ...
> > Topic already reifies a subject.
> > So by reifying a topic you are reifying the same subject that the original
> > topic had been reifying.
> > Thus in the following topics "a" and "c" are reifying the same subject.
> > If you want to talk about the <topic> resource itself, use topic "b".
> > 
> > <topic id="a"/>
> > <topic id="c">
> >   <subjectIdentity>
> >     <subjectIndicatorRef xlink:href="#a"/>
> >   </subjectIdentity>
> > </topic>
> > <topic id="b">
> >   <subjectIdentity>
> >     <resourceRef xlink:href="#a"/>
> >   </subjectIdentity>
> > </topic>
> > 
> > Did oversimplify your problem?
> 
> Yes, because what is at issue is to have a topic - that is, the actual 
> data structure that is held in a program and persisted in a topic map 
> data base - be the subject of conversation.  Many practical topic map 
> applications need to save metadata about their constructs - sometimes, 
> lots of data.  For example, what icon is to be used to represent a 
> particular topic type?  When was such-and-such a topic (not the subject 
> it stands for, but the actual topic) created and by who?  Who is allowed 
> to access it?
> 
> I would almost agree with Kal that "topic information item" would be 
> what is being referred to here, but I am unsure whether an information 
> item still exists after a topic map has been ingested and processed.  I 
> see it as an artifact of the description of the ingest process.
> 
> If I had any doubts as to the utility of such meta data - and how 
> essential it is that the data be persisted and shared - they would have 
> been dispelled by the 1-week training course in Metis that I just went 
> through.
> 
> And of course, it is not just topics but any "information item" in a 
> topic map that may need to become the subject of conversation.
> 
> It would be a grave error to fail to provide for this.  Of course, as I 
> have already posted, I regard this as a non-issue.  Just create a PSI 
> for the concept of "topic information item in a processed topic map", 
> create a topic of that type, and use an occurrence to point to the id 
> (or locator) of the target topic.  Presto!  Problem solved.
> 

Thats exactly the kind of thing I meant by a "best practice" - and I
tend to agree with you that there is probably no need for this to be
addressed in the data model. However those working on TM*L might want to
consider the problem of referring to topic information items as it might
impinge on their work...in which case then the data model does need to
be tuned to support this.

Cheers,

Kal

-- 
Kal Ahmed <kal@techquila.com>
techquila