Re: Arcs

On 21/10/10 13:41, Xuân Baldauf wrote:
[...]
> Benjamin Bock has proposed arcs in Prague 2009 and I have used arcs
> without problems in a project. My definition of "arc" is as follows:
>
>     An arc denotes an association perspective. An association
>     perspective with respect to a particular association is a
>     perspective as seen
>
>        1. by a player of such an association
>        2. though this association
>        3. towards another player of this association.
>
>     Arcs are independent of the particular association, hence they are
>     independent of the particular players of an association. Hence, each
>     particular arc is defined by
>
>        1. ("from") the type of the role of the local player
>        2. ("type") the type of the association
>        3. ("to") the type of the role of the remote player
>
> (This definition may be incomplete, as more-than-2-ary-associations may
> not be well-covered, but I haven't needed arcs for these types of
> associations so far.)

The semantics of Topic Maps are only meant to provide the underpinning
graph and not pollute that which is being modeled, so I'll assume you
mean this proposal as a Topic Map document, or perhaps as what we've
in the past called a "PSI Set": a published set of well-defined term
and relation types identified by URI, rather than as part of Topic Maps
semantics proper.

 From your description 'arcs' sound suspiciously like what are also
called 'triples' (binary relations) or 'tuples' (n-ary relations) in
many existing modeling languages. You've limited the relation to directed
binary, which would be suitable for building bipartite graphs[1]. I'll
assume that this is a design decision, a deliberate constraint on the
graph -- it is a significant that limits its application. In any case,
I don't see an innovation here (and perhaps none is intended).

If, for example, you wanted sentence-scoped relations of the form
'subject-predicate-object' this would be a suitable relation type. If
that is not the intent, since 'arc' has an existing meaning in basic
graph theory (e.g., Tuttle [2]) to describe a directed edge, I would
suggest using a different name; it's close enough to be confusing
unless one is deliberately adopting those semantics. This seems
awfully low in the model chain (at the level of a mathematical
formalism), certainly too low to be used for linguistic modeling.

[...]
> So, what does the community think about arcs? I think it is time to
> create some community modeling standards, and something like arcs should
> be parts of these standards.

An 'arc' expresses a specific ontological commitment, and while it is
demonstrably useful (as is any similar association type or set of
association types comprising a well-defined, bounded ontology), I don't
see that it represents anything significant nor anything that deserves
standardisation. It's just basic graph theory, and as such it might
even be a bit misleading to people who don't understand that
distinction, e.g., a lot of people using RDF inappropriately.

We as a community have for many years discussed creating some kind of
forum for the publication of "community modeling standards", and while
this may be considered useful it also may lead to a proliferation of
new projects creating new definitions for old terms and relations.
We don't really need that, and in fact that should be considered as
counter-productive. What would be more valuable would be to take
existing "ontologies" (for lack of a better term) and provide a means
(best practices) for assigning URIs to the existing terms and relations.

I've myself done that dozens of times, and back in 2002 or so was
trying to convince the Bureau International des Poids et Mesures to
assign URIs to all of their SI terms [3] so that we could use an
existing standard for measurement units. While my contact at the time
suggested they were looking into doing that, they haven't yet, but that
to me is a much more productive approach. If we as a community had some
kind of document that set out to explain why this was important it
might help lobby organisations to identify their terms as interoperable
'nouns' (see [4]).

It's kind of a boy thing to want to lay claim to semantic territories,
like planting a little flag on the moon, and I admit to having been
a boy in the past. Nowadays I'd rather the owners plant their own flags.
Then everyone can use those identifiers canonically, rather than demand
that people use *our* namespaces as wrappers for their semantics. That
would promote usage across all communities.

Murray

[1] http://en.wikipedia.org/wiki/Bipartite_graph
[2] p.3-4, 131. Graph Theory, W.T. Tuttle, 1984. Addison-Wesley, Menlo Park
[3] http://www1.bipm.org/en/si/base_units/
[4] http://tomayko.com/writings/rest-to-my-wife
...........................................................................
Murray Altheim <murray10 at altheim dot com>                       ===  = =
http://www.altheim.com/murray/                                     = =  ===
SGML Grease Monkey, Banjo Player, Wantanabe Zen Monk               = =  = =

       Boundless wind and moon - the eye within eyes,
       Inexhaustible heaven and earth - the light beyond light,
       The willow dark, the flower bright - ten thousand houses,
       Knock at any door - there's one who will respond.
                                       -- The Blue Cliff Record
_______________________________________________
topicmapmail mailing list
[email protected]
http://www.infoloom.com/mailman/listinfo/topicmapmail