Re: [topicmaps-comment] RE: [sc34wg3] Re: PMTM4 and XTM Layer 1.0

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

[Steven R. Newcomb]

> Maybe this will help to clarify:
>
> I'm *not* saying that people should *ignore* the
> assertions that are made about a subject when they're
> trying to understand that subject.  I'm saying that if
> there is no subject indicator, and therefore the
> assertions made about a subject are all that there is
> to indicate what the subject is, and if those
> assertions are made by more than one author (as will
> likely be the case in any sort of collaborative
> environment), then it's very likely that the exact
> nature of that subject will become multiple and/or
> ambiguous.  When that happens, the whole Topic Maps
> paradigm becomes unreliable and essentially useless.
> The paradigm simply doesn't work unless there is
> exactly one utterly changeless subject for every topic.
>
> Worse: in a collaborative environment that doesn't
> require collaborators to provide good subject
> indicators, the paradigm can *seem to be working* when
> it's really *not working*, and that can be very
> dangerous.
>
> I'm just making a plea that providing every topic with
> a good subject indicator (i.e., a subject indicator
> that is compelling, precise, and unambiguous) is an
> essential policy when attempting to produce useful
> topic maps by means of collaborative processes.
>

Steve, I see what you are saying and I agree - sort of.  I think that
getting such pristine and unambiguous subject indicators acoss a
collaboration ( or across an uncollaborated web) will be harder than you
seem to think.  I'm not ready to concede that topic maps will be as fragile
in this regard as you are suggesting.  Even such an apparently understood
concept such as "property" can be up for grabs - people on one of the RDF
lists have just been talking about differences between the DAML and RDF
notions of "property", just as an example.

Well, probably the best thing is for lots of people to get good working
experience with useful topicmaps.  Then we will start to see the real issues
emerging.  Which brings me to a different question.

Are there any substantial topic maps out there that have some real depth to
them?  By that, I mean depth of hierarchy, or even better, that they would
be more of a network than a hierarchy.  Most of the maps I have seen do not
contain many associations linking things, one to another to another.

For example, Lar's big topic map on free XML software.  You have one or two
top level categories, then a topic for software, then one or two
associations to get its author, etc, then you are done, nowhere to go.  This
is very appropriate for that kind of map, but I'm interested in looking at
maps with a lot more depth, with cycles, and so on.  We need maps like this
to really explore navigation - this leads to that leads to that other thing
leads to ...

Anyone?

Cheers,

Tom P