Re: [xtm-wg] A challenge on "the graph"

["W. Eliot Kimber" <eliot@isogen.com>]

Murray Altheim wrote:
> 
> "W. Eliot Kimber" wrote:
> 
> What I was trying to express (perhaps not so clearly as I'd like) is that
> I think we're looking for something extremely simple, but formal. While
> it is possible to describe a set of constraints (in prose) on a graphical
> language, what is really needed here are constraints at all but a formal
> modelling method. I'd be happy with circles and arcs if we could find
> one that was in common use, didn't require our reinvention or redefinition
> (as is the case with UML, where that is our only option unless we adopt
> somebody else's, like ObjectDomain's), itself had a formal definition,
> and was clear and easy enough to understand that it wouldn't be readily
> misinterpreted by half the planet.

My point is that there cannot exist a graphical representation that is
both simple and capable of expressing graphically all the graph
constraints you might need to express. Thus, assuming that part of the
model will be graphical (which I think is a prerequisite for
understandability by a non-specialist audience), then there will be, by
necessity, some set of constraints or rules that are defined as
mathematical expressions in some expression syntax. If this is the case,
then these expressions could be bound to UML-based diagrams *as easily
as* any other suitable graphical representation.

I can guarantee that in the domain of implementors, UML will be the
graphical notation that has the widest recognition--no other graphical
language will have the same penetration to that audience.

If there is some existing mathematical notation for writing down rules
about graphs, this language can, for certain, be applied to UML models
that define the nodes and arcs in the graph, assuming that this notation
provides some way to refer to nodes by name (class name or instance ID).
I cannot imagine a useful notation that wouldn't have this property.
 
> Perhaps UML is the ticket. Perhaps the XTM data model can be described
> formally in UML. As I've said previously, I remain to be convinced. I'd
> ideally like to find something that was more specific to graph theory,
> try that out and if if doesn't fit, we then know that graphs aren't the
> best way to express topic map concepts. Since we (at least in theory)
> have some good graph experts in the room it's at least not a bad start.

It's not a question of whether or not the XTM data model *can* be
defined formally in UML--it absolutely can and I can prove it by
existence of similar models that I have personally defined with
mathematical precision. The question is whether or not UML is *the best
choice*. It may or may not be. But it is absolutely *a candidate*, and,
I would think, a very strong candidate.

I'm championing UML here because I think it is the best fit for the
task:

- It has wide recognition and acceptance in industry
- It is a clear and simple visual language (for the uses that we need of
it--we would only be using a minimal subset of the total UML language)
- It has a built-in facilities for binding in formal statements about
the model
- There are well-defined modeling methodologies built on it (e.g.,
Catalysis)
- It is an industry standard of serious authority (OMG)
- There are free tools you can use to work with it (ObjectDomain, some
open-source systems)

As an implementor of the eventual model, UML will be the best possible
form of definition for me to use because it integrates directly into my
normal engineering tool set. Anything else, including property sets,
will require significant effort to map to UML as the starting point for
my implementation work. I assume that it would be the same for most
other implementors.

Cheers,

E.
-- 
. . . . . . . . . . . . . . . . . . . . . . . .

W. Eliot Kimber | Lead Brain

1016 La Posada Dr. | Suite 240 | Austin TX  78752
    T 512.656.4139 |  F 512.419.1860 | eliot@isogen.com

w w w . d a t a c h a n n e l . c o m

------------------------ Yahoo! Groups Sponsor ---------------------~-~>
Secure your servers with 128-bit SSL encryption!
Grab your copy of VeriSign's FREE Guide,
"Securing Your Web site for Business." Get it now!
http://us.click.yahoo.com/4cW4jC/e.WCAA/bT0EAA/2n6YlB/TM
---------------------------------------------------------------------_->

To Post a message, send it to:   xtm-wg@eGroups.com

To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/