Ed,
Just a little
further elaboration on some of your points. Comments are interleaved and
labeled [PC-1] to [PC-5]:
Pat
Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx
> -----Original Message-----
> From: Ed Barkmeyer [mailto:edbark@xxxxxxxx]
> Sent: Tuesday, January 06, 2009 3:19 PM
> To: Patrick Cassidy
> Cc: '[ontolog-forum] '
> Subject: Re: [ontolog-forum] Next steps in using
ontologies as
> standards
>
> Patrick Cassidy wrote:
> > The problem with letting the "market"
determine standards is that
> there has
> > to be an effective "market", with
multiple candidates, and multiple
> users,
> > for it to work.
>
> Yes.
>
> > In the case of a foundation ontology, there
have been publicly
> > available candidates for over 6 years, but as
yet there are
> few
> > users (applications, anyone?) and nothing
remotely resembling a
> "market" has
> > developed.
>
> I understand you are talking about a "standard
upper ontology", and
> not an ontological version of the Oxford Dictionary
of English.
>
> With my standards experience hat on, I would say
that 'nothing
> remotely resembling a "market"' after 6
years translates to academic shelfware.
> If no one is using it, one of the following must be
true:
> - it does not effectively support any
practice that is (currently)
> perceived to need support; or
> - it is not being used by the people and
tools engaged in the
> practice it was intended to support. Ignorance
is a possible
> explanation; not-invented-here is another; and there
are many other
> motivations for standards avoidance.
>
[[PC-1]] there are other possibilities that I think are
actually the problem.
(1) people are exploring the use of these ontologies, and
actually using them internally, but nothing dramatic has been thus far released
for public inspection, and probably nothing of general interest has been
developed in spite of ongoing efforts. See point (2)
(2) because a foundation ontology is as complex as the
basic vocabulary of a human language, it is time-consuming to learn how to use
to best effect, and without an existing dramatic application, efforts to use it
are hesitant and starved for funds. I have seen that problem
personally. This is the chicken-and-egg problem - the first seriously
interesting publicly visible application will take a long time to incubate,
because of its complexity and low level of funding.
(3) the lack of broad agreement on which of the
foundation ontology candidates will become the widely used standard inhibits
commitment of significant funding to any one of them. No one wants
to risk going down a dead-end.
(4) Developing an application - even database integration
via ontology - is much more costly than developing a foundation ontology that
can handle a local problem. As in (2), lack of prior visible successful
applications leads to hesitancy and very low funding for ontology -related
projects. In the government agencies where I worked the first question
asked is "is this technique being used in a program of record?" to
which the answer was, when I heard it asked a couple of year ago "no, not
to my knowledge". The response then is - "we aren't doing research
here, and we aren't going to be the first to break in a new
technology". Very short-term horizons for locally funded project
inhibits use of a new technology.
The third point can be addressed by the [process I
suggest of creating a large user community by funding participation of a large
number of developers of the standard. The killer app will still have to
come after the creation of the common foundation ontology.
The short-term horizon effect means that funding will be
very slow in coming for ontology applications, and generally inadequate to
create the impressive applications that will encourage further use.
Getting the breakthrough visible use of ontology in a significant application
will, I feel sure, be accelerated if there is some open-source foundation
ontology that is developed by a large group and therefore has a greater
likelihood of persisting and developing into a standard than any of the current
FO's. If it is founded on semantic primitives, and therefore has the
smallest possible size for its function, it is even more likely to be adopted
because of its greater ease of use than other FO's.
> I'm personally not sure which of those is the case
here. Cyc and SUMO
> were intended to support ontology development, but
they don't use the
> languages in vogue, and that alone would be reason
2, seen as reason 1.
>
> The "common practice" of ontology
development is based on RDF and OWL
> and CLIF-like FOL languages. Adam Pease's opus
falls into the last
> category at least, but there was no standard
language in that category
> until 2007, and every use of SUMO has required
twiddling with the
> formulation to get it into the tool-of-choice.
>
> But even if you can get the model into your tool,
there is the
> question of whether the chosen formulation of a
concept enables the
> tool to do useful reasoning. So support for
the practice requires not
> only formulation in a language the reasoning tool
can understand, but
> also a pattern of axioms that works well with the
reasoning algorithms
> used by the tool.
>
[[PC-2]] The choice of some reasoner to use with an FO
may well be very important, as you suggest. The FO should be in a Common
Logic conformant language, and should be susceptible to interpretation by a FOL
reasoner like Vampire or Prover9, or the Ontology Works system. But there
are variations in the way certain structures are handled (such as forall-exists
axioms) and these variations may have serious effect on the results. I
would expect that a project to create a common FO would include a component to
choose and tune a reasoner to work well with the FO. A some point
third-party vendors may develop better reasoners that use the same FO, and that
would be all to the better, but I agree that it will be important to make sure
that some reasoner works well with the FO.
> > This should give us a clue that we are dealing
with a technology
> > that is not simplistically analogous to the
ones we are accustomed to.
>
> That is another possible explanation, i.e., that
this is uncharted
> territory, and I agree it is also valid.
>
> It is uncharted territory in the sense that it is
not common practice
> to standardize a model and stop there. ISO and
other organizations
> have published "standard reference models"
for various things, but
> their primary purpose is to organize a
programme-of-work for creating
> useful standards that work together.
Otherwise, we normally
> standardize models as part of standardizing a
practice.
>
> As indicated above, however, the form and structure
of upper ontology
> models is critical to their being effective in
support of the ontology
> development practices. Because there is not
wide agreement on
> reasoning algorithms, the target practice is not so
clear, and thus
> the effectiveness of the standard models may be much
narrower than
> intended.
>
> > I suggest that, in dealing with a truly
powerful and potentially
> > revolutionary technology that is aimed at
supporting the replication
> of the
> > thinking function of humans, we keep an open
mind about what
> approaches are
> > likely to work.
>
> Yes! And while we are keeping our minds and
options open, it will be
> really hard to standardize models that work with unspecified
reasoning
> processes. Taking this to its "logical
conclusion", when we figure
> out how human reasoning works and we can replicate
it well enough
> (Asimov's positronic brain), then we can make the
fundamental
> reference ontologies that work well with that
process.
>
[[PC-3]] Well, in the absence of positronics I think that
some variant of an FOL reasoner will have to be used, but that variant, I
agree, needs to be thoroughly tested, and perhaps modified, to accommodate the
intended interpretations of syntactical structures in the FO. As an
example, The way that "forall-exists" statements are handled needs to
be carefully defined, and there may be (I hope there are) more than one way to
do that. Vampire just skolemizes any instances that are implied but not
explicitly defined in the ontology, with the result that the reasoning module
produces dense and almost uninterpretable inference traces (for example, from
SUMO) containing multiple skolemized individuals that can be very
difficult (for me) to interpret and recognize throughout the course of a
trace. On the other hand, the Ontology Works system allow two
possibilities: IC hard and IC soft existentials. IC hard just stops input
and refuses data when an implied individual is not explicitly defined. IC
soft outputs a warning and continues data input. Neither kind of
forall-exists statement is used at query time, the IC-hard existentials are
just used like database "integrity constraints" - an elegant solution
to the complexity introduced by such axioms. But I can envision the use
of alternative or additional treatments of forall-exists axioms, such as using
some of them (specifically labeled) only during certain reasoning processes, or
allowing some of them (specifically labeled) to be used to automatically
*create* instances at data input time, but not at query time.
> Until then, we have to make models that work with
the majority of the
> tools on the "market". If there is
no emerging market, there is no
> reason to make a standard for it. If there is
no majority, it is not
> possible to make a useful standard. But if
there are significant
> pluralities and identifiable RoI, we can make
multiple such standards.
>
> (Unfortunately, the Cyc experience was that the cost
of the upper
> ontology was generally higher than the return,
partly because of a
> continuing evolution that created inconsistencies
with prior models.
> And the early experience with SUMO had similar
problems. And rightly
> or
> wrongly, that experience has tainted and stunted
further efforts.)
>
> > Sure, past experience must be consulted, but
when new
> > technologies are being developed, over-rigid
analogies with previous
> > experience may well be more misleading than
helpful. I respectfully
> suggest
> > that prior work on information standards is
just not relevant to this
> issue.
>
> And I respectfully suggest that Santayana was
right. Wilful ignorance
> of the past is the bane of the information
technology industry. The
> adolescent mentality of software engineers is
unbelievable: "We have a
> new technology; the experience of our forebears
couldn't possibly be
> relevant."
>
[[PC-4]] IT is not "ignorance" to observe past
experience and recognize where false analogies can be misleading. The
best past experience for basing a foundation ontology comes from use of limited
defining vocabularies in some dictionaries. Because the words used in the
dictionary definitions are labels for concepts, it is reasonable to infer that
there is also a limited set of defining concepts (and ontological
representations of those concepts) that would allow ontological description of
an unlimited number of terms, concepts, or real-world entities in many
domains. But the experience of past standards efforts are relevant for
the foundation ontology only at a stage after there is agreement on the
inventory of primitive concepts and ontological representations thereof that
will suffice to form the "conceptual defining vocabulary". At
that point, making a standard of the initial foundation ontology may well
benefit from prior standards development experience. Perhaps there were
cases where there was a need to develop agreement on some set of primitive data
elements (the Java language?) that eventually developed into a standard.
Tales of how that happened might have some relevance, but I suspect that all
prior standards had such limited semantic scope that we are truly in uncharted
waters in developing an FO as a set of semantic primitive components.
> > I might also suggest that the current economic
situation might give
> one
> > pause in relying exclusively on the
"market" to solve issues.
>
> Or in postulating a project that requires massive
funding. ;-)
>
[[PC-5]] I don't think that $30 million over 3 years
qualifies as "massive" in comparison with the costs of lack of
semantic interoperability (100 billion per year), or to the costs (government
and private) of prior and ongoing efforts to address the semantic
interoperability problem by less effective methods. (How much has been spent on
Cyc? NIEM? NLP? Data Warehouses? OWL and its accoutrements?).
Daschle suggests that he will spend 10 billion per year for several years to
create a system for exchange of patient health information. $30 million
is about 0.1 percent of that figure. Very reasonable, considering that
all of the computer-interpretable specification of the data could be based on
the common foundation ontology.
> -Ed
>
> --
> Edward J.
Barkmeyer
Email: edbark@xxxxxxxx
> National Institute of Standards & Technology
> Manufacturing Systems Integration Division
> 100 Bureau Drive, Stop
8263
Tel: +1 301-975-3528
> Gaithersburg, MD
20899-8263
FAX: +1 301-975-4694
>
> "The opinions expressed above do not reflect
consensus of NIST,
> and have not been reviewed by any
Government authority."