SUO: Re: Unanimous Consent
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
Jay,
Problem statement.
1. What are the different types of ontology projects
that are covered by our current scope and purpose?
2. What are the criteria that are appropriate
to each of the different ontology projects?
NB. Until a better term comes along, I am using "project"
somewhat in the way that people speak of cultural projects
or existential projects -- broad, compelling, if slightly
vague intimations of something that needs to be done.
Here is a narrative about one sort of ontology project,
the aims, criteria, and working assumptions of which
I am acquainted with, and feel like I understand:
I once got sold on the project of building software bridges between
qualitative and quantitative research. For example, in many areas
of clinical practice, medical anthropology, and public health one
has "practitioner-scientist models" where people accumulate lots
of free-floating informal hunches and qualitative impressions in
their on-the-job settings, that they then need to follow up with
hard data gathering, quantitatively measurable constructs, and
the usual battery of statistical methods. A lot of practical
savvy never gets widely distributed, and a lot of benighted
mythology never gets tested, for the lack of good ways to
refine "personal knowledge" into scientific truth.
It still seems to me that properly designed lexical and logical resources
ought to provide us with some of the the planks for building that bridge.
At first strike, it sounds like this ought to involve an integration of
research oriented and common sense ontologies. But there has seemed to
arise one insurmountable obstacle after another in trying to do this.
Just by way of focusing on a concrete illustration, take the word "event".
Formalizing the concept of "event" for a research oriented ontology does
not require any discusssion on our part. Those discussions were carried
out somewhere between the days of powdered-wig-wearing-high-rollers and
the days of manurial comparisons. To get the standard axioms, one goes
to a standard reference book and copies them into one's knowledge base.
The only question is whether one's favorite ontology prover is up to
the snuff of proving whatever theorems need to be proved thereon.
There can be no compromise with these criteria.
The research market simply will not bear it.
So if there is to be an integration with
nontechnical language and methodology,
it must be an augmentation of these
basics and not their overwriting.
I have gotten used to the idea that there is another sort of ontology project,
but since I do not get the cogency of it, it seems like its definition and its
criteria of validity would have to come from the critical self-examination of
those whose project it is. All I know at present is that the obvious course
that I suggested above for formalizing the concept "event" is probably the
course of last resort from the standpoint of this alternative project.
That is what I mean by radical differences in working criteria for acceptance.
Similar disjunctions of approach and acceptability could be observed
for several other dimensions of diversity among ontological projects,
for example, the "already been chewed" vs. the "knowledge soup" brands,
that is, those who expect full-fledged axiom systems from the outset
vs. those who would gel their knowledge chunks out of a semiotic sol.
Jon Awbrey
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
Jay Halcomb wrote:
>
> Precisely so, Jon. I think that we've many of us said these similar things
> at one time or another, and we always return to them when a proposal is made
> (recall the discussion about the CycL language when that proposal was made).
> That is why I think that developing clearer acceptance criteria, upfront,
> for specifying these various targets is important, when it comes to working
> documents for the group. Specifically, developing specification criteria
> for terminologies, languages, and logic(s). I would hope the IFF folks
> should have some specific thoughts about this.
>
> Jay
>
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o