ontolog-forum
[Top] [All Lists]

[ontolog-forum] Fw: Fw: Semantic Webshortcomings [wasRe: ANN: GoodRelati

To: <ontolog-forum@xxxxxxxxxxxxxxxx>, <cassidy@xxxxxxxxx>
From: "Sean Barker" <sean.barker@xxxxxxxxxxxxx>
Date: Tue, 19 Aug 2008 20:26:47 +0100
Message-id: <000501c90231$85c6a150$0100a8c0@PackardDesk>

 Pat    (01)

The STEP project started in the mid-eighties, or perhaps
slightly earlier, and was born out of frustration with the existing CAD
standard of the era, IGES. The early outcomes of the project were the
data modelling language EXPRESS, which has been used independently of
STEP, and the fundamental concepts. The first standard to come out in
1995 was AP 203, configuration controlled 3D mechanical geometry. Much
of this is basic geometry, but perhaps the key set of concepts to come
out of this is the six fold division of product, product-version,
version-view, view-property, property-representation, and
representation-presentation. This division is absolutely fundamental for
coherent management of complex products, and the underlying ideas have
lead on to massive changes in the way design and manufacturing is
understood. One fundamental result is that virtually the only thing you can 
say unequivicooly
about a product is its name - everything else depends on the view you take, 
and generally you
need several views which are structurally and conceptually incompatible.    (02)

Subsequently, a whole series of other AP's (application domains
specific exchange standards) have been developed, and I will only
comment on two.    (03)

The first, the PDM schema, was a reharmonisation of AP 203 and
AP 214, which had diverged because of the different ambitions of the US
aerospace and European automotive industries. It describes the
information needed to co-ordinate configuration change. The test
implementation of this was a four way exchange - I don't know the costs on 
this, but the
implementation involved a team of three or four people in each of four
companies over a period of about a year. One of the key things that it 
demonstrated was that
the semantics of the terms used depended on the business processes of
the people doing the exchange, and that successful exchange is in part
about process harmonisation.    (04)

The second, PLCS, covers the support world. The first couple of
years of work involved some thirty or so subject matter experts agreeing
a standardised model of process flows for the support business. The next
stage involved some ten data modellers over two years, most of whom had
significant industry experience, and understood the SME terminology.
On average, the people were about half time on the project. It
also involved re-engineering STEP to add some fundamental concepts which
are not part of the design world, and didn't come up in the
manufacturing APs. One way of looking at PLCS is that it reifies the implied 
metaphysics
of the product development world - arguments about the nature of identity 
took several months to resolve.    (05)

The costs of $500,000,000 covers development of detailed
exchange standards and implementations of those standards, so more than
just the core of fundamental concepts. However, without implementation
and testing, the standard would have been wrong, the expectations for it
would have been wrong, and the funding would have dried up years ago.    (06)

Four things that need to be factored in to your costs are:
1) International co-operation at least doubles measurable time-scales
and costs (as opposed to the quadrupling time scales and costs when
national standards are created and then need harmonizing).
2) Not all contributors to development will have the same level of
expertise and commitment, and initially this can be a drag on the
project.
3) There will be at least one breakaway group who think they can do it
better, and may well point this out vociferously when it comes to
getting something agreed.
4) The vendors will go off and do their own thing, negating all your
work, unless you have some serious political clout.    (07)

The problem is that once all these things are taken into
account, the payback time is longer that the three years that is used
for many industrial funding decisions. Previous e-mails have pointed out
why there is no hope of getting the academics to do anything useful,
because their criteria are academic brownie points not business cost
reduction. And without credible demonstrators throughout the development
process (which need to be factored in to costs) funding dries up.    (08)

And STEP - 25 years on, and the big business and the defence
ministries are still funding its development. The biggest problem is,
where it is used, it works - however, there is very little wow factor in
showing a chief exec two identical computer screens, which is all we are
trying to do.    (09)

>From the perspective of "fundamental concepts" can I add this observation 
from my STEP
experience: There is no such concept as 'property' - 'property' is only a 
language game that
allows you to associate a descriptive value with what is being described. 
There are product properties
such as colour -"this car is green"; there are activity properties such as 
duration -
"this task took 2 hours"; and there are properties that are common to both, 
such as cost -
"this car cost $20000", "This task costs $200". What they have in common is 
phraseology, a language
game we both understand. If there is no such concept as property, can you 
coherently claim that
everything can be described in a few thousand fundamental concepts, since 
what distinguishes them is a language game?    (010)

Sean Barker
Bristol    (011)


From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Patrick
Cassidy
Sent: 18 August 2008 22:46
To: '[ontolog-forum] '
Subject: Re: [ontolog-forum] Fw: Semantic Webshortcomings [wasRe: ANN:
GoodRelations - The Web Ontologyfor E-Commerce]    (012)

Sean,
  It's hard for me to imagine why a *basic* data model (5000-1000
elements) would cost that much.  Do you have any references that
describe the history of the project, how it was constructed, and with a
cost tag on each phase?
I have a feeling that a lot more than just construction of the basic
model is included in that number.  Cyc has about 600 person-years, which
should not be over $60 million, and a lot of that was specialized, well
outside the boundaries of the foundation ontology.
  The five million dollar cost is the cost of demonstrating agreement
(not necessarily universal) on the most basic 2000 elements, which would
demonstrate the feasibility and utility of extending the ontology to a
full foundation ontology, and providing the proposal how to proceed.  To
include several demonstration applications and a usable natural-language
interface (converting definitions in a controlled vocabulary to the
logical form, and finding the nearest existing element(s) to the
proposed new element, I am estimating at another 10 to 20 million.    (013)

  A great deal depends on how one plans to proceed.  After 13 years of
discussion about "upper ontologies" and large already-existing
inventories of logically-defined basic concepts from the current most
well-known ones, I think we can devise a process that is a lot more
efficient than starting from scratch.    (014)

  Pat    (015)

Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx    (016)


> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
> bounces@xxxxxxxxxxxxxxxx] On Behalf Of Sean Barker
> Sent: Monday, August 18, 2008 2:14 PM
> To: ontolog-forum@xxxxxxxxxxxxxxxx; cassidy@xxxxxxxxx
> Subject: [ontolog-forum] Fw: Semantic Web shortcomings [wasRe: ANN:
> GoodRelations - The Web Ontology for E-Commerce]
>
>
>
>  Pat,
> The investment to date in the STEP data model has been estimated at
> over $500,000,000, and that is limited to the fairly narrow field of
> Europe/US Defence/Aerospace/Automotive product data. $5,000,000 for a
> foundation ontology sounds like a gross underestimate.
>
>
> Sean Barker
> Bristol
>
> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
> [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Patrick
> Cassidy
> Sent: 18 August 2008 15:54
> To: '[ontolog-forum] '
> Subject: Re: [ontolog-forum] Semantic Web shortcomings [wasRe: ANN:
> GoodRelations - The Web Ontology for E-Commerce]
>
>
>                *** WARNING ***
>
> This mail has originated outside your organization, either from an
> external partner or the Global Internet.
>      Keep this in mind if you answer this message.
>
> All those points about how traditional standards are developed are
> valid, but a foundation ontology for semantic interoperability is not
> like a traditional standard.  It's orders of magnitude more complex.
> There are already several possible starting points - Cyc is the most
> highly developed, but has been badly hindered by its commercial origin    (017)

> and continuing lack of full openness in development.  The fact that
> nothing has gained traction in over ten years should be an indicator
> that a new initiative is needed.  To me the obvious thing to try is to    (018)

> get together a large group of ontology developers and users and find a    (019)

> common *basis* (the foundation ontology) for creating logical
> representations of the meanings in all of the concepts that that group    (020)

> is interested in.  Such a project would cost over 5 million dollars,
> and such a project has never been funded - even though the benefits of    (021)

> success would dwarf the cost of development.  When that tactic has
> been tried and fails to get a large and growing user community, then
> and only then would I look for alternative methods that would be
> invariably more costly, slower and less likely to achieve the optimal
> solution.
> Whatever is developed by the starting project can evolve and adapt
> just as well - probably better, having been carefully thought out at
> the basic level - as anything mashed together by a less organized
project.
> Acceptance in the commercial field would follow after non-commercial
> development and applications have shown its usefulness.  Possibly the
> closest analogy would be the Linux operating system, where the core
> was developed by one person and is maintained by a tightly organized
group.
> But even Linux is simpler than a foundation ontology.
>    One consideration that seems to be ignored by those who are waiting    (022)

> for some standard to evolve from an unorganized collaborative process
> is that there is a very large cost in lost opportunity for every day
> the adoption of the standard is delayed.  The cost of just the lack of    (023)

> interoperability of relational databases in the US has been estimated
> at over 100 billion dollars per year.  The current lost opportunity
> cost for one hour would pay for a project to try to reach such an
> agreement.
> The benefits are so enormous that I think that *every* plausible
> tactic to achieve agreement on a foundation ontology should be funded.    (024)

> This notion doesn't seem to have been accepted yet by any funding
agency.
> Waiting for something to somehow appear by a process that has never
> produced any comparably complex artifact is not in my estimation a
> cost-effective tactic.
>
> Pat
>
> Patrick Cassidy
> MICRA, Inc.
> 908-561-3416
> cell: 908-565-4053
> cassidy@xxxxxxxxx
>
>
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Subscribe/Config: 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 Post:
> mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>    (025)


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config:
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 Post:
mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (026)




_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: 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 Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (027)

<Prev in Thread] Current Thread [Next in Thread>