Re: [topicmaps-comment] TMs & XTM [Was: skills to create topic maps]

[Mary Nishikawa <nisikawa@fuchinobe.skk.slb.com>]

Hi Robert,

Thanks for your reply.

At 02:27 PM 12/28/01 +1000, Robert Barta wrote:
>On Wed, Dec 19, 2001 at 03:02:44PM +0900, Mary Nishikawa wrote:
> > I am at the exploratory stage now with Topic Maps, .......................
> > .................................................. and have been
> > wondering  about  how to get a huge classification system with 
> thousands of
> > topics into topic maps (I would guess that it would not be practical  to
> > work with only one TM for this number of topics).
>
>If those topics have a structure, why not exploit it?

I can see now where a classification in the database could be mapped to 
topics and occurrences; The associations would need to be worked on later, 
I think, but at least there would be  a starting point.


>What we do here is to put everything into databases as soon as there
>is some inherent structure in the data. This database is then converted
>on-the-fly into a TM representation.
>
> > I would also like to work with XTM, but I haven't done so yet, so I am
> > interesting in learning how you are creating your maps using LTM.   For
> > those using XTM to create topic maps,  I see the advantages of having
> > everyone using the same dtd for interchangeability,
>
>You only need a DTD/XML schema if you are using XML to constrain its 
>(syntactic)
>structure. XMLSpy is suitable for that although it will not warn you (or 
>does it?)
>if you have a typo in your resourceRef internal references. You will
>need a development environment to get this sort of quality assurance.

Yes, I actually did that this week. I am not interesting in building 
complex TMs by hand, so I think XML Spy will suit my purposes for now. It 
will flag non unique ids which is useful. I do not look at the text, but 
use their enhanced grid view. About the typos, you are right.
So, which environment do your recommend for checking this?

I am most interested in content analysis and any method that would help me 
model the data in a generalized way and then apply it to categories of 
topics. Unfortunately, I can't do this in XML Spy.

I am also planning to write XTM PSIs for the TM pubsubj TC. I will validate 
my PSI files with the xtm dtd. After that, which development environment do 
you recommend to do the checking?

> > ................................................. but the ability to use
> > an abstract element declaration for each type that can be used to design
> > one's own schema to create a customized topic map has advantages  that
> > shouldn't be missed (I have to say I am looking for  colleagues to
> > try  this approach out). Thinking about how to create the schema helps me
> > organize my information better, which is what I am most interested in.
>
>I seriously doubt (please correct/brutalize/... me in case) that an XML
>schema instance allows to express _semantical_ constraints on a map.

I did not go into all the details and I am still exploring.

> > Related to the original posting [skills to create topic maps]  if someone
> > were to take this approach, I think the TM creator would need to be quite
> > familiar with the W3C XML Schema structures and datatypes, 
> namespaces,  and
> > Xlink for a start
>
>My students here have almost no idea about all this, still some of them
>create valuable maps even as absolute beginners. We send them on a mission
>after blessing them with
>
>   http://topicmaps.bond.edu.au/tutorial1/
>   http://topicmaps.bond.edu.au/astma/astma.html

This is interesting. Thanks.  I think that there are many ways that TM will 
be created, used, etc. If someone is already familar with xsd schemas, they 
may want to take a look at the xml schema for TM if they are interesting in 
the standard.


> > ............................. Yet again, if we can figure out how to map
> > what is in the database into the TM...
>
>Perl is a precious friend here.

Do you have a perl conversion script  for a mapping (from what to what)?

Cheers,
Mary