Dear Matthew,
If you are basing your economic
projections on the learning curve, I certainly do agree that the learning curve
works well. For each doubling of experience, the typical costs savings is
about twenty to thirty percent for nearly any valid learning situation,
especially in software.
But that consideration is distinct from
the technology of ontologies. I can learn how to export and import data
using a variety of tools, methods, designs, and whatever variation I want to
try. The question I have is why does ontology do any better at that than
alternatives to ontology? What is brought to bear that is so economically
attractive about ontologized export/import compared to smart coding,
recognition of commonalities in practice, and the usual economies of scale?
Please keep trying to convince me; I just don’t
see a clear argument for ontologized export/import yet.
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Matthew West
Sent: Thursday, March 01, 2012
9:03 AM
To: '[ontolog-forum]
'
Subject: Re: [ontolog-forum] What goes into a Lexicon?
Dear Rich,
You are saying that, since it is too
expensive to do for 1+1 systems, lets do it for thousands? That’s
like saying lets lose a little money on each sale, but make it up in volume!
MW: No when you do
it for 100s of systems, the costs multiply, and the benefits of using an
ontology come into play. Consider a network of 100 systems with 10 interfaces
per system. Never mind the initial cost of building 10 interfaces rather than
one. That could easily creep up on you. If we assume that each of these systems
lasts 20 years before being replaced, then each year 5 systems will be
replaced, and 50 interfaces will need to be replaced (instead of 5 using an
ontology).
It’s a fairly
compelling case. But I am please if you cannot see it.
My experience is that if one DB is
programmed to export data to company X, then that same export will be
considered when a second company Y has to be fed data.
MW: No. They will be
the customer so they will tell you what format they want it in, unless you have
an industry level ontology that you have both signed up to.
The incremental cost of adding a new
export routine is only paid once,
MW: No. It will
change as soon as one of the customer’s systems change.
and the format of the export becomes the
standard.
MW: Standard for
whom? It is not the supplier that sets the standards.
There is no need to do it differently for
the next company, for the most part.
MW: You really are
an optimist aren’t you.
Regards
Matthew
West
Information
Junction
Tel: +44 1489
880185
Mobile: +44 750
3385279
Skype:
dr.matthew.west
matthew.west@xxxxxxxxxxxxxxxxxxxxxxxxx
http://www.informationjunction.co.uk/
http://www.matthew-west.org.uk/
This email
originates from Information Junction Ltd. Registered in England and
Wales No. 6632177.
Registered office:
2 Brookside, Meadow Way,
Letchworth Garden City, Hertfordshire,
SG6 3JE.
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
From:
ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Matthew West
Sent: Thursday, March 01, 2012
12:17 AM
To: '[ontolog-forum]
'
Subject: Re: [ontolog-forum] What goes into a Lexicon?
Dear Rich,
Thanks for the description of the example
nuclear reactor. But I still question the value of adding the task of
making an ontology, instead of just exporting data from a thirty year old
system without the ontology step.
MW: I probably would
not do it for just one system to one other system either. But how about for
some hundreds of systems to some hundreds of other systems, where each system
interfaces to on average 10 other systems. Oh, and it is not just one company
that you are doing this for, but a whole industry. Much of the problem is
supply chain related, where data is transferred e,g, from equipment suppliers
to design contractors, to construction contractors, to owner operators, to
maintenance contractors, to decommissioning contractors. Does each party in
that chain want to have a bespoke way of exchanging information with each other
party, or is it better to have one way to communicate?
How many times do
you want to solve the same problem before you try to look for a shared
solution?
In general, when I have to move data from
one system to another, I just move the SQL tables, columns, domains (where
compatible), views and (where necessary) stored procedures. Why would it
be useful to first define an ontology for that thirty year old system?
What benefits would the company get from adding that apparently unnecessary
task to the activity of moving data from one system to another? If there
is no benefit, no company would include it. So in your example, there has
to be some benefit above and beyond data transfer. Could you please explain
what benefit it might be to add the ontology task to taking data and/or
software from a thirty year old reactor? You must have something in mind,
but I am not following the rationale just yet.
MW: What the
ontology gives you is a shared language that each system can translate
into and out of, rather than having a distinct way to do it for each pair of
system and company.
Regards
Matthew
West
Information
Junction
Tel: +44 1489
880185
Mobile: +44 750
3385279
Skype:
dr.matthew.west
matthew.west@xxxxxxxxxxxxxxxxxxxxxxxxx
http://www.informationjunction.co.uk/
http://www.matthew-west.org.uk/
This email
originates from Information Junction Ltd. Registered in England and
Wales No. 6632177.
Registered office:
2 Brookside, Meadow Way,
Letchworth Garden City, Hertfordshire,
SG6 3JE.
Moving data from one RDBMS to another is
just not that hard. The problem is in understanding the data that people
have actually typed into the database over the last thirty years.
Different people put in different text descriptions all the time.
An ontology would have to enforce strong
data typing of columns to be able to do any inference, but the data typing is
just not there in thirty year old systems. Furthermore, adding that
strong type checking at the data entry point is, for most systems, not productive
use of labor, and in many cases, makes performance so sluggish as to impact the
systems fitness for use.
Thanks for keeping at it though; I really
am trying to find some value in compensation for the cost of constructing the
ontology before transferring data. I just don’t follow that
argument yet. It doesn’t fit my experience with systems, whether
legacy or new. Please continue to explain it.
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
From:
ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Matthew West
Sent: Wednesday, February 29, 2012
2:20 PM
To: '[ontolog-forum]
'
Subject: Re: [ontolog-forum] What goes into a Lexicon?
Dear Rich,
The problem is not
working on 30 year old code. That probably stopped years ago, and the software
would have been abandoned if it did not work, so it is of sufficient quality
doing something rather mundane, that it does not need working on. The problem
is how do you get the data out of this system and into the systems you need to
decommission and deconstruct your nuclear reactor? How would you know how to
interpret anything you could get out? The same would apply if you were going to
replace the software of course.
You are not trying
to tack ontologies on top. Unless you had developed an ontology 30 years ago,
you would not have captured the semantics of the system outside the heads of
those that did the development, so you would not have access to those
semantics. So you need to develop an ontology of a system so that at some
later date you can make use of the data in ways that were not anticipated when
the system was built.
Regards
Matthew
West
Information
Junction
Tel: +44 1489
880185
Mobile: +44 750
3385279
Skype:
dr.matthew.west
matthew.west@xxxxxxxxxxxxxxxxxxxxxxxxx
http://www.informationjunction.co.uk/
http://www.matthew-west.org.uk/
This email
originates from Information Junction Ltd. Registered in England and
Wales No. 6632177.
Registered office:
2 Brookside, Meadow Way,
Letchworth Garden City, Hertfordshire,
SG6 3JE.
From:
ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Rich Cooper
Sent: 29 February 2012 19:03
To: '[ontolog-forum]
'
Subject: Re: [ontolog-forum] What goes into a Lexicon?
Dear Matthew,
My advice to anyone working on programs
that were written thirty years ago is to find another job. The technology
is outdated, the tools have become much, much better, languages are more
expressive, and subsystems can be licensed far more effectively now. My
advice to managers who have a thirty year old software system of significant
size is to muddle along as best they can while building an entirely new
replacement using modern technology.
The only value in creaking along with
thirty year old technology is in hoping it will go away soon and be replaced by
something more functional.
In any case, the sunk cost of that 30 year
old project has no current value other than avoiding replacement costs.
So why try to tack ontologies on top of something with a very limited
lifespan? I see ontologies, if they have a place at all, as newly
emerging solutions to yet unidentified problems. Our concern should be to
identify exactly which kinds of problems can be solved with ontologies.
Only then will they have clear value.
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Matthew West
Sent: Wednesday, February 29, 2012
10:48 AM
To: '[ontolog-forum]
'
Subject: Re: [ontolog-forum] What goes into a Lexicon?
Dear Rich,
My experience in software development in
teams is that the vocabulary used is absolutely essential to the two
programmers discussing their current issue of interfacing with each
other. Whether other programmers use the same word or not isn’t
significant to them; they are not writing programs to be readable until
possibly after the said programs actually work. So the problem is already
solved before any ontology is used, dictated, or agreed to. Then
there’s time to adjust words to fit some manager’s choice of vocabulary,
but that is AFTER the problem of a working program has already been
solved.
And what about the
situation when program A was written 30 years ago to support a nuclear power
plant, the writer of which has since died, and the writer of the second
programme now has to write interfaces to programs needed to decommission that
nuclear power plant over the next 20 years.
Regards
Matthew
West
Information
Junction
Tel: +44 1489
880185
Mobile: +44 750
3385279
Skype: dr.matthew.west
matthew.west@xxxxxxxxxxxxxxxxxxxxxxxxx
http://www.informationjunction.co.uk/
http://www.matthew-west.org.uk/
This email
originates from Information Junction Ltd. Registered in England and
Wales No. 6632177.
Registered office:
2 Brookside, Meadow Way,
Letchworth Garden City, Hertfordshire,
SG6 3JE.