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

["Thomas B. Passin" <tpassin@home.com>]

[Lars Marius Garshol]


>
> That, and a text editor, is really all you need to get started on
> writing LTM topic maps. Ontopia's Omnigator can display LTM topic maps
> (provided you call them something.ltm) just like it displays XTM topic
> maps. If you want your topic map in XTM format you can just export it
> from the Omnigator using the export plug-in.
>
> | 2) What did you have before using LTM (e.g., a spreadsheet listing
> | all the topics and their associations)?
>
> Actually, I didn't have anything at all. I just started writing,
> extending and modifying the ontology as I went. Later, when I got more
> serious about the topic map, I bought Daniels and Bright's "The
> World's Writing Systems" and used that as the data source.
>
> | 3) What are the "sufficiently" technical prerequisites for the
> | author?
>
> Well, basically they need to understand the basic concepts of topic
> maps, and be sufficiently technical that they don't balk at the idea
> of authoring topic maps in textual form. I would say that anyone who
> can write HTML pages in a text editor can use LTM.
>

The main difficulty in hand-authoring topic maps is not syntax, nor the use
of xml vs. ltm, although compactness does help a person scanning a partially
completed map, and a compact syntax is easier to type.  The ISO flavor is
helpful here, being more compact and readable.  But these are not decisive
matters.

The key points are these, in my view:

1) Finding the ID values for previously defined topics, so they can be
referenced.  This is easy in a tiny map, but gets progressively harder  as
the map gets larger. Also, it gets harder and harder to remember which
topics you have already defined, and naming can get to be a problem.
Similarly, it gets harder and harder to decide on what scopes to create and
apply, and you start to need help in seeing which ones you have allready
assigned to a name or occurrence.

When the map gets bigger, you also find it harder to modify it - especially
deletions and corrections.

2) Deciding how to model various kinds of information, ie., occurrences vs
associations, and then being consistent about them.  This includes any
reification you may be tempted to do.

3) Creating your ontology and being very clear about when something is an
instance vs. being a subclass of another topic.

The first of these is a sort of mechanical thing, and having a computer
program can help a lot.  The other two are really about knowledge
engineering and take time and practice to master.  Your approach may depend
strongly on the software you expect to have available to display and process
your maps.

Compared to these issues,the matter of the exact syntax used is of little
importance, as long as you can arrrange to transform it to a standard like
XTM at some point.  You should always make sure you have some way to do so,
otherwise you won't be able to use anyone else's tools.  Lars ponted out one
way to do this.

Cheers,

Tom P