Pat, (01)
I'll start from the end of your last post. There is little point in repeatedly
referring to lost productivity on a grand scale when there is no one concerned
about it who has the resources to address it. The closest match would require a
call to 202-456-1414, ask for Obama. Good luck with that. (02)
My experiences with DOD affirm your comment in that they want an order of
magnitude advance from semantic technologies. But the hard truth is that the
more tightly a domain is focused, the more a relational DB solution will equal,
or better a semantic solution. At much less cost and difficulty. (03)
The DOD entities I've dealt with typically seek to integrate some number of
relational DB's, sometimes with unstructured data. But within a fairly tight
domain (yes, there is a range to the domain focus). (04)
Interoperability, while given much lip service, lags the commercial world. The
DOD XML Registry was going to be the solution...still waiting on that one. By
the way, where is the ontology repository and how capable is it? (05)
Fielding large scale, operational semantic systems is not yet feasible because
the supporting infrastructure is not in place. For example, look in the Sunday
paper job ads for oracle DB administrators. Then look for ontologist ads... (06)
Unfortunately, as it takes money to make money, success breeds success. Fielded
semantic systems that work will garner more funds for the application of those
technologies. (07)
It would be great if the federal government CIO announced a major drive to
develop and field semantic technologies. I'm not holding my breath and he is a
lot more worried about the next windows virus that decimates federal PC's. (08)
It will take a near gorilla campaign of small successful semantic projects time
to prove themselves while the supporting infrastructure matures. (09)
The same way that Apple makes sound loops available in garage bands for
neophytes to compose into music may be a model for non-ontologists to grab
pertinent ontologies to compose into domain ontologies. Then there will be a
customer pull for the techologies often discussed here. (010)
This was a hellofa long email to type on a blackberry. Good thing my flight was
delayed. (011)
Don (012)
----- Original Message -----
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
<ontolog-forum-bounces@xxxxxxxxxxxxxxxx>
To: '[ontolog-forum] ' <ontolog-forum@xxxxxxxxxxxxxxxx>
Sent: Fri Feb 20 17:09:34 2009
Subject: Re: [ontolog-forum] standard ontology (013)
Just one point here: (014)
[JS] > And if Pat seriously believes that it's necessary to adopt some
> upper ontology ASAP, I can't understand why he doesn't recommend
> Cyc instead of starting from scratch.
>
Yea gods, how many times do I need to repeat that we are not starting
from scratch, but using *everything* freely usable from the OpenCyc??? Come
on, read my postings for their content, not just to find lines to argue
with.
I have already detailed reasons why Cyc in its present form and with
its present management cannot serve as a common foundation ontology
acceptable to a broad community. It seems you haven't read my
point-by-point enumeration of the reasons. At a minimum, the common FO must
be maintained by an open consortium of users, not by a central profit-making
organization. And the point of the FO project is not to develop a
replacement for Cyc, which is likely to continue in its present form for the
foreseeable future, but to develop a community that uses some FO that can
translate from Cyc and any other ontology into each other, and to
demonstrate its usefulness by producing open-source applications and
utilities, especially an NL interface, to make it easier to use. Such a
community and such demo applications do not now exist for any foundation
ontology. I have said repeatedly that if there were a broadly
representative community of users that openly shared their applications and
utilities, any competent foundation ontology, including Cyc, could be used
as a starting point, only modified as required to support translation and
serve to allow the users' systems to transmit information to each other. (015)
One of the things we should have learned from the experience of Cyc and
other work on upper ontologies is that the foundation ontology needs to be
freely usable, but even that is not enough. An FO serves as a language to
support communication among computers, and every useful language is
maintained by the community of users to be sure it suits their evolving
purposes. Central control is too inflexible - what is needed is a process
that allows input from every source, but that also maintains a core
component that is logically consistent, because computer communication can
and should be highly accurate. That notion of process argues for a
completely open consortium of contributors, with a technical committee
responsible for maintaining the quality of the FO. (016)
Where Doug's original model failed is that he thought that he only had to
develop a competent ontology, and could then license it to users. The
problem, we have discovered, is that using a complex ontology to good effect
is just as complicated as building the ontology in the first place, and as a
result users with applications important enough to pay license fees, for
which the managers are convinced that the ontology will pay its way, will be
very slow in arriving. This says nothing about the usefulness or quality of
the ontology, and a lot about how much work is required to implement a
complex technology in a practical application. (017)
Why is it so hard to use an ontology to create impressive applications?
>From comments I have heard within the DoD, it appears that anything that
does not approach human-level understanding in sophistication will be almost
indistinguishable from the managerial view from ordinary RDB's or ordinary
applications, and there are a lot more people who know how to do RDB's and
traditional programming. Inertia is immense. Getting someone to put up
money to prove that the ontology will help their applications is very hard.
There are some projects, but they are confidential and cannot be talked
about. (018)
As an example, consider language understanding. A good language
understanding program that could read a text and answer questions, and also
hold a coherent conversation, even (as JS points out) at a five-year old
level might be impressive enough. But what would it take to create such a
program? One architecture I favor is to have an NLU program highly
modularized, so that individual modules can be improved and replaced to
evolve a more powerful system, with an unlimited number of people working on
the system, if it is open-source. But language understanding is complex,
and I believe that the complexity will be at a minimum that which would be
associated with a "lexical-syntactic expert" approach, in which many,
perhaps most individual words will have to be interpreted by "expert"
modules designed specifically for individual words, phrases, grammatical
constructions, and contexts. Reasonable fluency at a minimal level will
require a good understanding of at least 5,000 words. My best
guess-estimate of how long it would take for a programmer (perhaps in
consultation with an ontologist) to write and debug such an "expert" module
would be one or two weeks. Take the lower number, and we get 5,000
person-weeks. Double that to allow for the work required to manage the
interactions of such modules, result 10,000 person-weeks, or 200
person-years. I also suspect that *any* program that came close to
human-level intelligence in a practical application would take as much time
as that. That is the barrier to demonstrating why an ontology will be
worthwhile using in a practical application, as contrasted with traditional
IT approaches. (019)
So, if anyone is wondering why I don't just take Cyc and use it for a demo -
aside from the fact that it is still mostly proprietary, the answer is, I
feel certain that it is impossible for anyone person to create a demo that
is sufficiently more impressive than traditional IT methods so as to break
through the consciousness of managers who could fund real projects. (020)
We need a large community to create such a demo, and that is the main
function of the FO project, regardless of what ontology is eventually
adopted for that purpose. It just has to be an ontology accepted in common
and used to develop open-source applications that use each other's
information. The FO proposal is structured to create that community by
paying developers to join and contribute to it. (021)
And, Oh, while we have been debating this topic for the past month, the
country has blown at least 10 billion dollars in loss of efficiency due to
lack of semantic interoperability. If the cost/benefit estimates I have
suggested do not seems realistic, then will someone else provide a different
one? Even PatH's pessimistic 0.01% still results in a highly favorable
benefit/cost ratio. And no one has suggested an alternative except "well,
maybe the semantic web will eventually result in some level of
interoperability". Tick, tick, there goes another hundred billion with no
progress in that direction. (022)
Pat (023)
Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx (024)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (025)
sophistication will be almost
indistinguishable from the managerial view from ordinary RDB's or ordinary
applications, and there are a lot more people who know how to do RDB's and
traditional programming. Inertia is immense. Getting someone to put up
money to prove that the ontology will help their applications is very hard.
There are some projects, but they are confidential and cannot be talked
about. (026)
As an example, consider language understanding. A good language
understanding program that could read a text and answer questions, and also
hold a coherent conversation, even (as JS points out) at a five-year old
level might be impressive enough. But what would it take to create such a
program? One architecture I favor is to have an NLU program highly
modularized, so that individual modules can be improved and replaced to
evolve a more powerful system, with an unlimited number of people working on
the system, if it is open-source. But language understanding is complex,
and I believe that the complexity will be at a minimum that which would be
associated with a "lexical-syntactic expert" approach, in which many,
perhaps most individual words will have to be interpreted by "expert"
modules designed specifically for individual words, phrases, grammatical
constructions, and contexts. Reasonable fluency at a minimal level will
require a good un (027)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (028)
|