Re: SUO: a silly question about the new modular architecture
Adam and Matthew,
This is just one more reason why I believe the modular architecture
is important:
"West, Matthew R SITI-GREA-UK" wrote:
> Conceptually I think there is no difference between
> having two standardised concepts/axiom sets
> and choosing between them and
> one standardised one and one non standardised one
> and choosing to use the non-standardised one.
>
> There is a formal (standards) difference, but I expect
> most extensions/changes to SUO to come about initially
> in this way, and it seems healthy to me provided the
> limits of what is standardised are known.
As many of the recent discussions on SUO list have shown, some parts
of SUMO are fairly stable and noncontroversial (e.g., nobody seems
to argue about the axioms for well-understood mathematical theories).
But other topics, such as organizations, raise enormous numbers of
divergent opinions.
That is why I believe it is impossible to certify a monolitic
ontology as a single "standard". Instead, it is more appropriate
to certify a framework for presenting a modular collection of
theories, each of which has its own certification procedure.
It is very common in standards circles to provide a "registry" for
various kinds of specifications, which may have different levels
of certification. But I believe that a lattice-based or IFF-based
framework has an important advantage over traditional registries:
it provides a method for combining ontologies to generate new
combinations of the modules upon request.
Each module in the framework could have its own level of
certification. Some of them would be much more dependable than
others, and users could be told which ones have been tested, by
whom, and to what level of certification.
John Sowa