Perception and cognition

[John F Sowa]

The recent discussions with Alex in the thread about artificial neural
nets led me to update my slides on "The virtual reality of the mind".
But that file now has 112 slides.

So I deleted 93 slides to create a "short version" that states the
main themes about perception and cognition in just 19 slides:
http://www.jfsowa.com/talks/vrshort.pdf

Perception is the foundation for cognition, but cognition does
much more than perception. The cognitive functions are necessary
for all application systems, not just AI.

By the way, the title page of vrshort.pdf has a pointer to the long
version (112 slides). But I recommend vrshort for an overview.

John

[Dennis E. Wisnosky]

John:

Interesting that you conclude vrshort with the OODA loop.  Boyd was the father of the USAF F16 which was the last Fighter that actually needed a pilot.  F18 and F15 people might disagree with me.  His OODA loop was to train fighter pilots how to manage the beast.  Boyd's biography is interesting on many levels.

I taught the use of the OODA loop when I was lecturing on BPR and DoDAF.  Use this rather than belaboring a decision by requiring more data.  OO Decide and Act, then OO again.

We did the same while building the DoD Business Enterprise Architecture in 6 months after there previous team accomplished nothing useful in 4 years.

We are doing the same with FIBO now.

Best,

John

--
All contributions to this forum by its members are made under an open content license, open publication license, open source or free software license. Unless otherwise specified, all Ontolog Forum content shall be subject to the Creative Commons CC-BY-SA 4.0 License or its successors.
--- You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-forum+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Dennis E. Wisnosky 

Senior Advisor - Consultant, Enterprise Data Management Council

Founder, Wizdom Systems, Inc.

CTO-CA (ret), DoD Business Mission Area

DWiz C 630-240-6910

[henson graves]

Here is a paper given at an engineering conference in Cape Town in 2014 that suggests that the OODA loop is a good paradigm for use in "wicked problems"

---

From: ontolo...@googlegroups.com <ontolo...@googlegroups.com> on behalf of Dennis E. Wisnosky <den...@wisnosky.net>
Sent: Monday, June 26, 2017 10:56 AM
To: ontolo...@googlegroups.com
Subject: Re: [ontolog-forum] Perception and cognition

 

To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.

[Alex Shkotin]

John,

before to read, a funny illustration of the state of the art of AI perception and cognition https://youtu.be/QCqxOzKNFks

Alex

John

[Paola Di Maio]

Hay Alex

funny obnoxious and somewhat worrying video

Is that how intelligent people spend their time in Japan?

I have worked in ...er... knowledge based systems for some time. My conclusion so far is that, having overcome most technological barriers to development,

the current limit to developing intelligent systems is  indeed human stupidity

As scholars, and engineers, we constantly hit our own

cognitive limits as we hit the barrier of our own minds

So I am spending time working on understanding the mind, we need to bring in the limits of our perception/cognition 

into the AI picture before we can do something truly useful to humanity

:-)

Paola Di Maio about.me/paoladimaio

[Alex Shkotin]

Hi Paola,

2017-06-27 12:11 GMT+03:00 Paola Di Maio <paola....@gmail.com>:

Hay Alex

funny obnoxious and somewhat worrying video

Is that how intelligent people spend their time in Japan?

​NO. that was how Japan​

 

​kids do​.

Intelligent J-people are here https://youtu.be/-KxjVlaLBmk?list=PLDcDXmbn7uEnGb_oMkEgSg91w1aMsw5_f

​:-)​

​Alex​

[Alex Shkotin]

just in case we miss https://arxiv.org/abs/1706.05125

"Negotiations require complex communication and reasoning skills, but success is easy to measure, making this an interesting task for AI."

I did not read the text yet but it may be interesting.

Alex

2017-06-26 17:38 GMT+03:00 John F Sowa <so...@bestweb.net>:

John

[Paola Di Maio]

phew

much more interesting and reassuring, Alex

thanks-

The computation, signal processing and mechanical

plasticity combined allow great advances

Still, how to deploy and employ these capabilities rests with mind, I reiterate the case-

I know, AI scientists have not typically engaged with mind, but definitely its time to stop avoiding te core issues

Paola Di Maio about.me/paoladimaio

[John F Sowa]

Dennis, Henson, and Alex,

You raise some important issues. And by the way, the proposed ISO
standard for ontology ignores them.

As the slides for both vrmind and vrshort emphasize, the cognitive
cycle must address goals, intentions, planning, and value judgments.
Those issues are critical for software used in business, engineering,
communications, finance, government, law, medicine, and life. The
ontology for any data or software designed for those fields must
support them. But the proposed ISO standard ignores them.

Dennis
> [Boyd's] OODA loop was to train fighter pilots how to manage the beast.

Yes. I first learned about Boyd when I was giving a talk about Peirce's
cycle, and a DoD representative mentioned the similarity.

Henson

> the OODA loop is a good paradigm for use in "wicked problems"

Yes. Your article shows Boyd's later version of the loop. The original
version (for fighter pilots) was intended for cycles that take less than
a second to traverse. The longer cycle for "wicked problems" may take
minutes to days to years to traverse.

Excerpt from Henson's paper, wicked.pdf:
> In the realm of commercial applications, OODA Loop architectures are
> now main stream where large data stores are accessed and billions of
> events are processed. These software applications use inference from
> historical patterns, reason to gather further information and form
> conclusions that are then acted upon: alerts can be sent, processes
> can be triggered, and decisions can be made and acted upon.

I like these examples because they illustrate a range of important
problems for which the reasoning is much "deeper" than the so-called
"deep" neural nets. I admit that DNNs are valuable for some split-
second perception-based reasoning. But the longer cognition-based
cycles are necessary for a broad range of practical problems.

And even for split-second decisions, background knowledge and
advanced planning can be critical. They may take hours to set up
a strategic situation in which the tactics are executed in seconds.
We need cognitive cycles (OODA loops) at every time scale.

Alex

> a funny illustration of the state of the art of AI perception and
> cognition https://youtu.be/QCqxOzKNFks

I noticed that the Japanese are designing robots for SUMO wrestling.
This is apparently the lightweight competition. I wouldn't want
to be in the same room with the heavyweight competition.

John

[henson graves]

John, Dennis

Your observation about different time scales fits with my experience. The loop is a good paradigm for both cases. I was very familiar with the OODA loop from working at LM Aero. When working on some problems that involved very large data and almost millisecond decision making I realized that it was a very useful guide for design and analysis. Later I noticed that it worked for "wicked problems" were the time span may be years. A consequence of the loop is the development and refinement of an ontology as well as action taken. In all cases one wants to start with an initial ontology, such as one about what constitutes a target, or bank fraud, etc. Operation of the loop may modify the original ontology to be unrecognizable, but if you don't have a starting one it is hard to get anywhere by only data analysis, e.g, connectivity, reachability, etc.  The paper was intended as a tutorial  for engineers about how engineering has changed in recent years and why they need understanding of such things as ontology.

By the way what is the reference to the proposed ISO standard?

Henson

---

From: ontolo...@googlegroups.com <ontolo...@googlegroups.com> on behalf of John F Sowa <so...@bestweb.net>
Sent: Tuesday, June 27, 2017 8:20 AM

To: ontolo...@googlegroups.com
Subject: Re: [ontolog-forum] Perception and cognition

ロボット相撲 youtu.be Japanese Sumo Robots (Twitter: @id_r_mcgregor) Carefully edited takes from several years of footage. Source: https://www.youtube.com/watch?v=whqGS7tqa_M http...

I noticed that the Japanese are designing robots for SUMO wrestling.
This is apparently the lightweight competition.  I wouldn't want
to be in the same room with the heavyweight competition.

John

--
All contributions to this forum by its members are made under an open content license, open publication license, open source or free software license. Unless otherwise specified, all Ontolog Forum content shall be subject to the Creative Commons CC-BY-SA 4.0 License or its successors.
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

Google Groups groups.google.com Google Groups allows you to create and participate in online forums and email-based groups with a rich experience for community conversations.

[John F Sowa]

On 6/27/2017 12:08 PM, henson graves wrote:
> By the way what is the reference to the proposed ISO standard?

I'm sorry for not answering sooner, but I had to dig out the references,
and I was also distracted by some discussions in the univalent thread.

For a draft of Part 1, with comments by Matthew West, see
http://isotc.iso.org/livelink/livelink?func=ll&objId=18868196&objAction=Open

For a draft of Part 2, the BFO ontology, see
http://isotc.iso.org/livelink/livelink?func=ll&objId=19048488&objAction=Open

I mostly agree with Matthew's comments. My main criticism of the
current draft is that it says nothing about the complex issues
of knowledge sharing and semantic interoperability that have been
analyzed and debated for the past half century. For a summary of
those issues with 100+ references, see http://www.jfsowa.com/ikl .

For ontology, The most serious omission is the lack of any mention
of microtheories and their use in developing large ontologies.
Microtheories for ontologies were introduced by Cyc in 1991. Since
then, the largest ontologies have been designed as combinations of
microtheories. But Part 1 ignores that work.

The attached file, bfo_cat.jpg, is a tree of the BFO top-level
categories (Figure 1 in Part 2). It consists of a few philosophical
options, about which there is no consensus among philosophers:
continuant/occurrent; dependent/independent entities...

Some philosophers use those terms, many don't, and others criticize
them for various reasons. Therefore, BFO should not be considered
a consensus view among philosophers. That is not sufficient reason
for rejecting BFO completely, but it is a strong reason for not
giving BFO a privileged position.

Unfortunately, Part I begins with advertising hype that suggests
BFO is the foundation for the research on the human genome. In fact,
scientists in physics, chemistry, biology... never use the BFO terms
to state or describe their research data or theories.

In the physical sciences, everything is a process (occurrent).
An object (continuant) is a process that changes so slowly that it
can be recognized at repeated encounters. There is a continuum,
not a dichotomy.

Next, the terms 'dependent' and 'independent' do not characterize
anything in the physical sciences. They are linguistic distinctions
that characterize certain terms (such as adjectives or nominalized
adjectives) from the nouns they apply to. But every one of those
distinctions can be "explained away" in non-linguistic terms --
i.e., by a paraphrase that does not mention the language.

Finally, the distinction between material and immaterial should be
replaced by the terms 'physical' and 'mathematical'. Everything in
BFO that is characterized as immaterial comes from some mathematical
theory about geometry, mereology, topology, etc. With today's computer
technology, virtual reality (VR) is a far more precise and detailed
mathematical system for analyzing and describing phenomena in the
physical sciences -- which include biology and molecular biology.

In summary, there is nothing special about BFO that makes it more
appropriate than any other ontology, such as CYC, DOLCE, SUMO, or
COLORE, for specifying anything in science. The claim that the human
genome ontology uses any BFO terms is an accidental result of the
fact that Barry Smith and his colleagues wrote those definitions.

Claim: Any definition in the genome ontology that uses BFO could
be rewritten with the BFO terms replaced by English words that have
traditionally been used in biology and molecular biology -- and the
rewritten versions would be just as precise and probably clearer
than any version that uses BFO terms.

In some of his writings, Barry Smith quoted favorable comments by
some of the biologists. I'm sure that Barry and his colleagues used
the same kind of techniques as any knowledge engineer: ask probing
questions in a Socratic dialogue. That's useful, but there's nothing
in BFO that makes it more useful than any other ontology.

My claim is supported by a backhanded compliment by one of the
biologists: "BFO did not get in the way." In other words, BFO
was irrelevant: equivalent clarifications could have been achieved
without any special terminology from BFO.

Recommendation for the next draft of the proposed ISO standard:

1. In Part 1, remove *all* mention of BFO or any other specific
ontology. Use only terminology that is common to three or
more sources (of which BFO may be one).

2. Emphasize the use of microtheories in Part 1 and explain
how a large ontology can be constructed by combining
multiple microtheories. Define a top-level ontology as
a microtheory that does not use terms defined in some
other microtheory.

3. Add terminology and requirements to Part 1 from developers
such as Matthew West who had designed, developed, and used
large ontologies for industrial applications.

4. Design Part 2 as a collection of microtheories from various
sources. The BFO top-level is small enough to be a microtheory.
Other microtheories may be added to Part 1 from any source.

5. Ontologies that are too big to be included in Part 2 may
contribute microtheories that are sufficiently precise and
widely used. Cyc, for example, has over 6,000 microtheories,
many of which would be candidates for inclusion in Part 2.

John

[Matthew West]

Dear John,
I have been taking an interest in the emerging standard for Top Level Ontologies (I have become a member of the appropriate National Committee and am a UK nominated expert to the appropriate ISO/IEC subcommittee). So allow me to respond to some of your comments.

On 6/27/2017 12:08 PM, henson graves wrote:
> By the way what is the reference to the proposed ISO standard?

I'm sorry for not answering sooner, but I had to dig out the references, and I was also distracted by some discussions in the univalent thread.

For a draft of Part 1, with comments by Matthew West, see http://isotc.iso.org/livelink/livelink?func=ll&objId=18868196&objAction=Open

[MW>] There is a later draft, and another is due in a few weeks, probably with substantial changes.

For a draft of Part 2, the BFO ontology, see http://isotc.iso.org/livelink/livelink?func=ll&objId=19048488&objAction=Open

I mostly agree with Matthew's comments. My main criticism of the current draft is that it says nothing about the complex issues of knowledge sharing and semantic interoperability that have been analyzed and debated for the past half century. For a summary of those issues with 100+ references, see http://www.jfsowa.com/ikl .

[MW>] Indeed it does not, for the simple reason that is not the problem it is trying to solve. It is trying to do something much smaller, which is to set some standards for Top Level Ontologies that hopefully ontology developers will aspire to meet.

For ontology, The most serious omission is the lack of any mention of microtheories and their use in developing large ontologies.
Microtheories for ontologies were introduced by Cyc in 1991. Since then, the largest ontologies have been designed as combinations of microtheories. But Part 1 ignores that work.

[MW>] I agree microtheories are important, but are not the subject of this standard because it is a standard for Top Level Ontologies, which might encompass the top theories of an ontology or family of ontologies, but not microtheories.

The attached file, bfo_cat.jpg, is a tree of the BFO top-level categories (Figure 1 in Part 2). It consists of a few philosophical options, about which there is no consensus among philosophers:
continuant/occurrent; dependent/independent entities...

[MW>] I don't know any philosophical option about which there is consensus. This means it is very important in a top level ontology to state what options you are taking, and what sorts of things it allows you to talk about.

Some philosophers use those terms, many don't, and others criticize them for various reasons. Therefore, BFO should not be considered a consensus view among philosophers. That is not sufficient reason for rejecting BFO completely, but it is a strong reason for not giving BFO a privileged position.

[MW>] It is not presented that way. It is just one (the first) TLO to be presented as an ontology that meets the conformance requirements of Part 1. As such, even if you do not agree with the philosophical positions it takes, it acts as a useful example of what conformance looks like. Perhaps it will also act as a spur to those who hold other positions.

Unfortunately, Part I begins with advertising hype that suggests BFO is the foundation for the research on the human genome. In fact, scientists in physics, chemistry, biology... never use the BFO terms to state or describe their research data or theories.

[MW>] I'm assured that will change. I agree it is a mistake (see my comments).

In the physical sciences, everything is a process (occurrent).
An object (continuant) is a process that changes so slowly that it can be recognized at repeated encounters. There is a continuum, not a dichotomy.

Next, the terms 'dependent' and 'independent' do not characterize anything in the physical sciences. They are linguistic distinctions that characterize certain terms (such as adjectives or nominalized
adjectives) from the nouns they apply to. But every one of those distinctions can be "explained away" in non-linguistic terms -- i.e., by a paraphrase that does not mention the language.

Finally, the distinction between material and immaterial should be replaced by the terms 'physical' and 'mathematical'. Everything in BFO that is characterized as immaterial comes from some mathematical theory about geometry, mereology, topology, etc. With today's computer technology, virtual reality (VR) is a far more precise and detailed mathematical system for analyzing and describing phenomena in the physical sciences -- which include biology and molecular biology.

In summary, there is nothing special about BFO that makes it more appropriate than any other ontology, such as CYC, DOLCE, SUMO, or COLORE, for specifying anything in science. The claim that the human genome ontology uses any BFO terms is an accidental result of the fact that Barry Smith and his colleagues wrote those definitions.

[MW>] There is a subtlety here that I missed initially, and that is the distinction between a Top Level Ontology and an Upper Ontology. The latter is a rather broad term, the former intended to be rather more precise and limited. So a TLO is just the top level stuff that will be a part of most Upper Ontologies, like the ones you mention, but not all of them.

Claim: Any definition in the genome ontology that uses BFO could be rewritten with the BFO terms replaced by English words that have traditionally been used in biology and molecular biology -- and the rewritten versions would be just as precise and probably clearer than any version that uses BFO terms.

[MW>] Of course.

In some of his writings, Barry Smith quoted favorable comments by some of the biologists. I'm sure that Barry and his colleagues used the same kind of techniques as any knowledge engineer: ask probing questions in a Socratic dialogue. That's useful, but there's nothing in BFO that makes it more useful than any other ontology.

My claim is supported by a backhanded compliment by one of the
biologists: "BFO did not get in the way." In other words, BFO was irrelevant: equivalent clarifications could have been achieved without any special terminology from BFO.

[MW>] Even so that is better than most do.

Recommendation for the next draft of the proposed ISO standard:

1. In Part 1, remove *all* mention of BFO or any other specific
ontology. Use only terminology that is common to three or
more sources (of which BFO may be one).

[MW>] Likely will be done.

2. Emphasize the use of microtheories in Part 1 and explain
how a large ontology can be constructed by combining
multiple microtheories. Define a top-level ontology as
a microtheory that does not use terms defined in some
other microtheory.

[MW>] Not appropriate for the scope of the standard.

3. Add terminology and requirements to Part 1 from developers
such as Matthew West who had designed, developed, and used
large ontologies for industrial applications.

[MW>] I'm involved (so is Michael Gruninger).

4. Design Part 2 as a collection of microtheories from various
sources. The BFO top-level is small enough to be a microtheory.
Other microtheories may be added to Part 1 from any source.

[MW>] Not part of the scope. The parts after Part 1 will be other TLOs that wish to be standardised as alternatives to BFO. Alternatively, for existing standard TLOs, they can claim conformance to the standard.

5. Ontologies that are too big to be included in Part 2 may
contribute microtheories that are sufficiently precise and
widely used. Cyc, for example, has over 6,000 microtheories,
many of which would be candidates for inclusion in Part 2.

[MW>] Again, the aim is a standard for ontologies (well TLOs to be precise) not one giant standard ontology.

Regards

Matthew West
Information Junction
Mobile: +44 750 3385279
Skype: dr.matthew.west
matthe...@informationjunction.co.uk
http://www.informationjunction.co.uk/
https://www.matthew-west.org.uk/
This email originates from Information Junction Ltd. Registered in England and Wales No. 6632177.
Registered office: 28, Connemara Crescent, Whiteley, Fareham, Hampshire, PO15 7BE.

[John F Sowa]

Dear Matthew and Todd,

>> [JFS] My main criticism of the current draft is that it says

>> nothing about the complex issues of knowledge sharing and semantic
>> interoperability that have been analyzed and debated for the past
>> half century.
>

> [MW>] Indeed it does not, for the simple reason that is not the
> problem it is trying to solve. It is trying to do something much
> smaller, which is to set some standards for Top Level Ontologies
> that hopefully ontology developers will aspire to meet.

Unfortunately, that goal ignores the purpose of an ontology:
support interoperability among independently developed systems
-- *especially* the trillions of dollars of software that has
no explicit ontology of any kind.

This goal raises fundamental questions about how to design a TLO
that has that level of flexibility. There are several options:

1. Edict a single TLO, which serves as the hub of a family of
spokes. Each TLO is the hub of a silo that is incompatible
with every competing silo -- and with the multi-trillions
of dollars of legacy software.

2. Develop a theory for designing ontologies in a systematic way
that facilitates interoperability. This is the microtheory
hypothesis of CYC and most large modern ontologies.

3. Ignore the top levels and focus on interoperability at the
mid levels and lower levels. This is the basis for using
WordNet and other lexical resources to align the data at the
level of ordinary language. This method has been moderately
successful for legacy systems from the punched-era of the
1890s to nearly every system connected to the WWW today.

4. Develop mathematical methods for finding relationships among
independently developed systems. Examples include DOL and
other kinds of mathematical techniques. This method could
help #2 and #3 above. It might even relate some silos in #1.

Option #1 (the hub & spoke model) is a strategy to promote
incompatible silos. I believe that this strategy is the primary
reason why practical developers ignore ontology.

>> [JFS] Design Part 2 as a collection of microtheories from various

>> sources. The BFO top-level is small enough to be a microtheory.
>> Other microtheories may be added to Part 1 from any source.
>
> [MW>] Not part of the scope. The parts after Part 1 will be other
> TLOs that wish to be standardised as alternatives to BFO. Alternatively,
> for existing standard TLOs, they can claim conformance to the standard.

In short, the scope (copy below) implies that the standard will
consist of multiple competing TLOs, each of which is designed to
be the hub of a silo that is incompatible with every other silo.

Todd
> Will you be attending the IAOA Summer Institute on Upper Ontologies,
> to be held in Toronto, this August?

No. Just the thought of listening to people claim that their silo
is better than anybody else's silo provokes retchophobia. I have
better things to do.

John
__________________________________________________________________

An excerpt from the scope Part I: (Comment: silos by design.)

This International Standard focuses on ontologies to be used as
resources designed to support the interchange of information among
heterogeneous computer systems. It specifies a hub-and-spokes
architecture for ontology development and sets forth the requirements
an ontology shall satisfy if it is serve as hub in such an architecture.

The following are within the scope of this International Standard:

• Specification of how ontologies used for data retrieval, integration
and analysis can be combined into modular suites of mutually consistent
and non-redundant ontologies.

• Specification of the hub-and-spokes structure of such ontology suites.

• Specification of the role of definitions in a hub-and-spokes ontology
architecture.

• Specification of the requirements for an ontology to serve as hub in
such a structure.

[Matthew West]

Dear John,

-----Original Message-----
From: ontolo...@googlegroups.com [mailto:ontolo...@googlegroups.com] On Behalf Of John F Sowa
Sent: 05 July 2017 04:36
To: ontolo...@googlegroups.com

Subject: Re: [ontolog-forum] Proposed ISO standard for ontology

Dear Matthew and Todd,

>> [JFS] My main criticism of the current draft is that it says nothing
>> about the complex issues of knowledge sharing and semantic
>> interoperability that have been analyzed and debated for the past
>> half century.
>
> [MW>] Indeed it does not, for the simple reason that is not the
> problem it is trying to solve. It is trying to do something much
> smaller, which is to set some standards for Top Level Ontologies that
> hopefully ontology developers will aspire to meet.

Unfortunately, that goal ignores the purpose of an ontology:
support interoperability among independently developed systems
-- *especially* the trillions of dollars of software that has no explicit ontology of any kind.

This goal raises fundamental questions about how to design a TLO that has that level of flexibility. There are several options:

1. Edict a single TLO, which serves as the hub of a family of
spokes. Each TLO is the hub of a silo that is incompatible
with every competing silo -- and with the multi-trillions
of dollars of legacy software.

[MW>] This is the option chosen for this standard (others are welcome to develop standards for the other options). I agree silos are an issue, but if they are well enough defined at the top level, then mapping between them is not so hard.

2. Develop a theory for designing ontologies in a systematic way
that facilitates interoperability. This is the microtheory
hypothesis of CYC and most large modern ontologies.

[MW>] I thought Cyc was about capturing "common knowledge" which is quite a different thing.

3. Ignore the top levels and focus on interoperability at the
mid levels and lower levels. This is the basis for using
WordNet and other lexical resources to align the data at the
level of ordinary language. This method has been moderately
successful for legacy systems from the punched-era of the
1890s to nearly every system connected to the WWW today.

[MW>] That is a lot harder than mapping TLOs.

4. Develop mathematical methods for finding relationships among
independently developed systems. Examples include DOL and
other kinds of mathematical techniques. This method could
help #2 and #3 above. It might even relate some silos in #1.

Option #1 (the hub & spoke model) is a strategy to promote incompatible silos. I believe that this strategy is the primary reason why practical developers ignore ontology.

[MW>] I disagree. A more common problem is that ontologies are just not capable enough because most have not been developed with any TLO and therefore have local restricted views of things that are at the edge of their domain, but at the centre of another.

>> [JFS] Design Part 2 as a collection of microtheories from various
>> sources. The BFO top-level is small enough to be a microtheory.
>> Other microtheories may be added to Part 1 from any source.
>
> [MW>] Not part of the scope. The parts after Part 1 will be other TLOs
> that wish to be standardised as alternatives to BFO. Alternatively,
> for existing standard TLOs, they can claim conformance to the standard.

In short, the scope (copy below) implies that the standard will consist of multiple competing TLOs, each of which is designed to be the hub of a silo that is incompatible with every other silo.

[MW>] Yes. There just is not one TLO everyone will agree to. However, I think standardising a few so that they are comparable is a good step towards overcoming the differences among them.

[Todd Schneider]

John,

What I hope to gain from the IAOA meeting is better understanding
on what a 'theory of ontology' (in the context of information systems)
should encompass and what that may entail and possibly extract better
design principles. From a practical perspective l'll be pushing for
consensus
on common terminology (i.e., see the IAOA terminology list
http://iaoaedu.cs.uct.ac.za/pmwiki.php?n=IAOAEdu.TermList).

Also, the discussions should be entertaining.

Todd

[ontos Rob]

Re:Todd's comments (but related to the iso topic)
For a well-rounded understanding of theory of ontology in IS and possible design principles, there should be input and viewpoints from various sectors that have experience with ontologies, not simply academia.
But the meeting facilitators remain from academia sector. And if no one in the audience is from other sectors, it will be a partial view of a larger international platform that involves AI, Comp Sci, IS, etc.

The ontology community lacks consensus on various topics, including terminology. (and these things should be ironed-out *before* iso proposals are proposed).
And consensus requires participation, so again: if people--from various sectors and countries--do not participate then it won't be consensus. It'd definitely be helpful to have input from as many countries (particularly non-English speaking) as possible so that we can understand various conceptualizations, metaphysical views, interesting concepts, and approaches.
But if you look at the terminology link you provided, most are terms used by or drawn from bfo, and it is the only tlo listed at the bottom of the page.(But what about yamato, gfo, ufo, dolce, etc.?) This is similar to the iso proposal. Surely that (and the iso proposal) does not communicate an impartial approach that is fair to all tlos and all tlo development methods. 

[Mike Bennett]

I’ll be there and I am not from academia. I work in the financial sector and we are great believers in taking what comes out of the best of academia and putting it to work. I expect to bring some interesting challenges to the table.

 

Mike

 

From: 'ontos Rob' via ontolog-forum [mailto:ontolo...@googlegroups.com]
Sent: Wednesday, July 5, 2017 12:22 PM
To: ontolog-forum <ontolo...@googlegroups.com>
Cc: tjsch...@covad.net
Subject: Re: [ontolog-forum] Proposed ISO standard for ontology

 

Re:Todd's comments (but related to the iso topic)

--

[Michael Gruninger]

On 2017-07-05 12:41 PM, Mike Bennett wrote:

I’ll be there and I am not from academia. I work in the financial sector and we are great believers in taking what comes out of the best of academia and putting it to work. I expect to bring some interesting challenges to the table.

Indeed, Day 2 of the Summer Institute
http://ontologforum.org/index.php/SummerInstitute2017#Programme
is dedicated to the applications of upper ontologies (semantic integration, ontology design/reuse)
and discussions on whether they deliver the advertised benefits.

- michael

[henson graves]

In light of John and Matthew’s comments

One fears that any proposed upper ontology standard will at best not cause too much inconvenience and unnecessary expense. At worst, ...  John states “the purpose of an ontology: support interoperability among independently developed systems”. This seems to me to be correct. Simply picking an ontology doesn't address semantic interoperability.

As far as I can tell the only way to address semantic interoperability as well as the only way to certify an ontology is to understand its "ontological commitment". The only effective way to do that that I know of is to represent the ontology within a formal logic. Then the ontological commitment is what is true in any valid interpretation. This approach is feasible.  

The formal approach of an ontology as an axiomatic theory in some recognizable logic enables ontologies to be compared.  What one wants is the ability to unify, merge ontologies or find out where they are inconsistent, etc.

Henson

---

From: ontolo...@googlegroups.com <ontolo...@googlegroups.com> on behalf of John F Sowa <so...@bestweb.net>

Sent: Tuesday, July 4, 2017 10:35 PM
To: ontolo...@googlegroups.com

Subject: Re: [ontolog-forum] Proposed ISO standard for ontology

[Matthew West]

Dear Henson,
That's basically the approach I'm pushing for. We'll see what the next dtaft brings.
Regards
Matthew West

[ontos Rob]

Additionally, other concerns are as follows.

The current proposal is by a particular tlo, according to criteria of that tlo, so of course it will satisfy the proposed criteria...but this also means it is unfair to other tlos.

No other tlo's or developers were involved or had input from the beginning of proposal or in the design the criteria. No other tlo developers are participating or have authoritative editorial powers over the proposal. And there is no guarantee or commitment from other tlos that they will participate.
The proposal represents a bfo-centric view, not a view for all tlo's, let alone the international ontology-related community.

The potential negative consequences will therefore be more than merely inconvenience: it would result in an iso for a particular tlo, which would reduce the user-base of other tlos (=unfair), increase monopolization by that tlo, etc. The proposal is premature on a number of fronts.

[John F Sowa]

Dear Matthew and Henson,

This morning I realized that we could recommend a methodology that
combines Matthew's suggestions and mine: Communication among silos
(the hubs of all the spokes) can be based on commonalities among TLOs.
But microtheories play a central role in facilitating communication.

> [MW] I agree silos are an issue, but if they are well enough defined

> at the top level, then mapping between them is not so hard.

HG

> Simply picking an ontology doesn't address semantic interoperability.

> ... What one wants is the ability to unify, merge ontologies or

> find out where they are inconsistent, etc.

I agree with Matthew that having precise definitions is important.
But Part 1 should say that it's only the first step. Part 1 should
at least define the term 'microtheory', explain how they are used
in many ontologies, and state Henson's point about further operations
that microtheories can facilitate.

>> [JFS] Develop a theory for designing ontologies in a systematic

>> way that facilitates interoperability. This is the microtheory

>> hypothesis of Cyc and most large modern ontologies.

>
> [MW] I thought Cyc was about capturing "common knowledge" which
> is quite a different thing.

Every ontology represents "common knowledge" about something.
Even for a narrow domain, the subject matter may have an open-ended
variety of connections and relations with knowledge in other domains.

>> [JFS] WordNet and other lexical resources [are used] to align

>> the data at the level of ordinary language. This method has been
>> moderately successful for legacy systems from the punched-era of
>> the 1890s to nearly every system connected to the WWW today.
>
> [MW>] That is a lot harder than mapping TLOs.

On the contrary, this has been the *primary* method that has been used
informally for over a century. It's the basis for systems analysis
and knowledge engineering by humans and by computational methods for
NLP, data mining, data analytics, and systems such as IBM Watson.

> [MW>] There just is not one TLO everyone will agree to. However,

> I think standardising a few so that they are comparable is a good
> step towards overcoming the differences among them.

I agree that it's a good first step. But the method of finding and
using common microtheories is the next step. In effect, the human
programmers who write software to link independently developed systems
discovered *informal* microtheories. They should be formalized.

An important use for microtheories: By including a "wrapper" based
on an appropriate microtheory, a legacy system can interoperate
(share data) with systems based on formal ontologies.

To illustrate the methods, I'll compare the BFO tree (bfo_cat.jpg in
my previous note) to the top-level "crystal" from the KR ontology in
_Knowledge Representation_ (Sowa 2000). (See the attached KR_cat.gif.)

BFO and KR share some distinctions, but with some differences:

1. BFO puts the Continuant/Occurrent distinction at the top, and
KR puts it at the third level. In the KR book, I also say that
the Continuant/Occurrent distinction is not as fundamental as the
others because everything in physics is a process, and objects at
the most fundamental level are just slowly changing processes.

2. KR puts Physical/Abstract second, and BFO puts the equivalent
Material/Immaterial distinction under Independent-Continuant.
As a result, BFO has no place for six KR terms: Schema, Script,
Description, History, Reason, Purpose.

3. The three-way KR distinction of Independent/Relative/Mediating
is limited to a two-way dichotomy of Independent/Dependent in BFO.
That leaves no room for Structure, Situation, Reason, Purpose.

4. BFO also has some terms that are not included in the KR top level.
Most of them are about space, time, parts, and boundaries. In KR,
these issues are not ignored, but they are represented at lower
levels by replaceable microtheories with different options.

There is much more to be said about these issues, but microtheories
play a central role in all of them. A collection of reusable
microtheories can be much more valuable and *sharable* than the
original TLOs they were designed for.

John

[Pat Hayes]

> On Jul 5, 2017, at 7:38 AM, Todd Schneider <tjsch...@covad.net> wrote:
>
> John,
>
> What I hope to gain from the IAOA meeting is better understanding
> on what a 'theory of ontology' (in the context of information systems)
> should encompass and what that may entail and possibly extract better
> design principles.

That sounds eminently reasonable and a laudable goal. But…

> From a practical perspective l'll be pushing for consensus
> on common terminology (i.e., see the IAOA terminology list
> http://iaoaedu.cs.uct.ac.za/pmwiki.php?n=IAOAEdu.TermList).

…this is a disaster. Both the selection of terminology to standardize and the definitions offered are completely idiosyncratic, in some cases to the point of seeming wilfully obscure, and are completely influenced and dominated by the OBO philosophical tradition, ignoring all other uses of words like “class". To standardize on these would be simply to create another useless silo which will be ignored by some and enthusiastically treated as a gospel by others.

Pat Hayes

[Todd Schneider]

Pat,

I can understand you comment about 'consensus on terminology',
but in the context of the IAOA term list and its intent, I think another
look may be in order (i.e., I may have failed to properly explain myself).

The original intent of the list of terms was to collect those (natural
language)
terms deemed useful to understanding some of the background of the various
disciplines that contribute to ontology and that appear in many source
materials
(relating to ontology) and their more common definitions.

The assumption was that many terms do have different definitions and
being able
to provide these variations (and something of their context) would be
helpful in
promulgating a better understanding of ontology and its underpinnings.

At present the IAOA term list is undergoing revision to weed out less
relevant
entries (e.g., BFO, Common Logic, OWL, etc.).

The consensus I was (ambiguously) referring to was on the terms themselves
and possibly the various definitions that should be included. Not consensus
on a single definition (per term).

Todd

P.S. The best uses I've found from foundational ontologies (aka Upper
ontologies)
are the explanations, motivations, and reasoning behind the
various decisions
made in their creation.

In application, those explanations have been more flexible and
interoperable
the ontologies themselves.

[henson graves]

I applaud what appears to be progress in this discussion.

I have reservations about using commonalities as the methodology for communicating between silos, as I shudder to think about the time that could be spent sorting out commonalities among TLOs. John’s examples being to the point, e.g.,  Continuant/Occurrent vs an object (continuant) is a process that changes so slowly that it can be recognized at repeated encounters.  His other examples deserve serious study.  

The progress in the discussion, as I see it, comes by taking ontologies (microtheories) as first class citizens (Chris Menzel and others) with their semantic integration as the key issue to be addressed. Maybe someday some microtheory can be promoted to be a TLO if it plays its cards right, and sheds its more constraining parts.  

Regarding definitions of ontology terms in natural language, this should suffice for ontologies to classify pizza toppings, but it is not sufficient for engineering ontologies. Natural language descriptions in this arena can always be interpreted in slightly different ways, which leads to interoperability problems, sometimes with very expensive resolutions. This issue is at the heart of many problems of cost over runs, schedule delays, and poor products.  I have about 25 years’ experience in this arena.

In my opinion, science and engineering ontologies need to be, and can be, formalized, using logics with lots of graphics such as found in UML. The practitioners in these areas can take the formalized version as the authoritative source.

Henson

 

---

From: ontolo...@googlegroups.com <ontolo...@googlegroups.com> on behalf of John F Sowa <so...@bestweb.net>

Sent: Wednesday, July 5, 2017 2:10 PM

To: ontolo...@googlegroups.com
Subject: Re: [ontolog-forum] Proposed ISO standard for ontology

[David Price]

FWIW I agree 100% with Pat wrt the terms list. I had a quick look at a few there are mostly really terrible (i.e. they would make the current situation worse).

Look at the definition of “Ontology” for a prime example that will do nothing but add confusion:

Ontology: An ONTOLOGY is a representational artifact, comprising a taxonomy as proper part, whose representational units are intended to designate some combination of universals, defined classes, and certain relations between them

No consensus outside BFO would ever be reached on that (i.e. none of that text would survive review by anyone outside the BFO community).

Best stop that exercise before any more time is wasted, and rather spend time on other fruitful activities.

If for some reason people choose to continue, then every noun must have an adjective to specify context … so “BFO Ontology”, not “Ontology”. No point in saying “Ontology” and then list 10 completely conflicting definitions … that adds nothing wrt wider consensus or clarity. That must follow through into the definitions too. For example, the "BFO Ontology" definition depends on very specific definitions of the terms it uses and I’m sure other of less-than-useful definitions would also appear for those terms in the list. How would anyone not deeply involved even start to parse such definitions without a clear context adjective used everywhere, all the time?

I’m not 100% sure of the intended audience, but I would certainly *never* point an enterprise customer I was trying to convince of the value of a semantic approach to such a term list (even specified as I suggested) as it would provide nothing but ammunition for the naysayers.

Cheers,
David

UK +44 7788 561308
US +1 336 283 0606

[Matthew West]

Dear John,

Dear Matthew and Henson,

This morning I realized that we could recommend a methodology that combines Matthew's suggestions and mine: Communication among silos (the hubs of all the spokes) can be based on commonalities among TLOs.

[MW>] Commonalities are relatively easy, it is the differences that are challenging, so mapping a spatio-temporal extent to an endurant (something with temporal parts to something without them). I think it can be done mind you.

But microtheories play a central role in facilitating communication.

[MW>] There will certainly be common mapping patterns that can be derived from a mapping between TLOs.

> [MW] I agree silos are an issue, but if they are well enough defined
> at the top level, then mapping between them is not so hard.

HG
> Simply picking an ontology doesn't address semantic interoperability.
> ... What one wants is the ability to unify, merge ontologies or find
> out where they are inconsistent, etc.

I agree with Matthew that having precise definitions is important.
But Part 1 should say that it's only the first step. Part 1 should at least define the term 'microtheory', explain how they are used in many ontologies, and state Henson's point about further operations that microtheories can facilitate.

[MW>] Part 1 is trying to do one simple thing - set some standards for a TLO. Obviously there is much more to do, but that will need to wait for later standards.

>> [JFS] Develop a theory for designing ontologies in a systematic way
>> that facilitates interoperability. This is the microtheory
>> hypothesis of Cyc and most large modern ontologies.
>
> [MW] I thought Cyc was about capturing "common knowledge" which is
> quite a different thing.

Every ontology represents "common knowledge" about something.

[MW>] I meant heuristics as opposed to science. (Not that science is excluded).

Even for a narrow domain, the subject matter may have an open-ended variety of connections and relations with knowledge in other domains.

[MW>] Of course. Having a TLO for a range of domain ontologies is just one step towards interoperability.

>> [JFS] WordNet and other lexical resources [are used] to align the
>> data at the level of ordinary language. This method has been
>> moderately successful for legacy systems from the punched-era of the
>> 1890s to nearly every system connected to the WWW today.
>
> [MW>] That is a lot harder than mapping TLOs.

On the contrary, this has been the *primary* method that has been used informally for over a century.

[MW>] Yes, it has been the primary method, that does not mean it is good. When you look inside large organizations you find that they are often managed without the right information because of the cost and time it takes to gather reconcile and integrate it.

> [MW>] There just is not one TLO everyone will agree to. However, I
> think standardising a few so that they are comparable is a good step
> towards overcoming the differences among them.

I agree that it's a good first step. But the method of finding and using common microtheories is the next step. In effect, the human programmers who write software to link independently developed systems discovered *informal* microtheories. They should be formalized.

[MW>] Possibly. But if so that will be another standard.

[Matthew West]

Dear Rob,

 

 

Replies in-line below...

On Wednesday, July 5, 2017 at 8:10:13 PM UTC+1, John F Sowa wrote:

Dear Matthew and Henson,

> [MW>] There just is not one TLO everyone will agree to. However,
> I think standardising a few so that they are comparable is a good
> step towards overcoming the differences among them.

The current proposal does not standardise a few, but has always aimed at only one.

[MW>] That is not true. The standard is open ended as to the number of TLOs that can be standardised or claim conformance.

No other tlo is participating in the design of the criteria in this proposal or in general.

[MW>] Well I am, and I have at least one TLO I might seek to either claim conformance for or standardise here.

[John F Sowa]

Dear Matthew,

I basically agree with your comments. We might quibble about
some of the details and strategy, but there are more important
issues to discuss.

Just one important issue:

>>> [JFS] WordNet and other lexical resources [are used] to align the
>>> data at the level of ordinary language. This method has been
>>> moderately successful for legacy systems from the punched-era of
>>> the 1890s to nearly every system connected to the WWW today.
>>
>> [MW>] That is a lot harder than mapping TLOs.
>
> On the contrary, this has been the *primary* method that has been
> used informally for over a century.
>
> [MW>] Yes, it has been the primary method, that does not mean it is
> good. When you look inside large organizations you find that they
> are often managed without the right information because of the cost
> and time it takes to gather reconcile and integrate it.

I agree that we need something better and that good ontologies can
be very helpful -- even for designing the "wrappers" that enable
legacy systems to interoperate with the latest and greatest.

But the documentation for every system, old or new, starts with
and ends with natural languages. That includes everything from
the original proposal, the requirements documents, the design
documents, the user manuals, the comments in the code, the help
facilities, the tutorials, and all the emails at every step
along the way.

Furthermore, no matter how precisely you define any term with
any version of logic, the people who enter the data and use the
results will *never* see those precise definitions -- even if
they could, they wouldn't.

Fact: NLs won't go away until people go away.

John

[Chris Partridge]

FWIW I agree with David and Pat - that this really needs quite a lot of work to make it usable.

>>> To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-forum+unsubscribe@googlegroups.com.

>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
>
> --
> All contributions to this forum by its members are made under an open content license, open publication license, open source or free software license. Unless otherwise specified, all Ontolog Forum content shall be subject to the Creative Commons CC-BY-SA 4.0 License or its successors.
> --- You received this message because you are subscribed to the Google Groups "ontolog-forum" group.

> To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-forum+unsubscribe@googlegroups.com.

> For more options, visit https://groups.google.com/d/optout.

--
All contributions to this forum by its members are made under an open content license, open publication license, open source or free software license. Unless otherwise specified, all Ontolog Forum content shall be subject to the Creative Commons CC-BY-SA 4.0 License or its successors.
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-forum+unsubscribe@googlegroups.com.

[Todd Schneider]

David,

The current study and practice of ontology is multi-disciplinary. Each
discipline and
sub-discipline brings their own understanding and definitions. This
multi-disciplinary
aspect of ontology makes it a fascinating field and also difficult.

As I mentioned earlier the intent of the Term list is for education:
Provide a source for
the various terms and definitions that have use in the (involved)
disciplines and contribute
to a broader understanding.

The current state of the list and definitions is not what we would like
it to be and are
in the process of culling the list and reviewing the definitions. We'll
make use of your
suggestion.

We certainly encourage contributions which should include a source
reference. The source
reference is intended to provide the context for the definition.
Contributions can be sent
to iaoa-ed...@ovgu.de. Better yet join the IAOA Education committee.

There's a wonderful practice I learned from the Open Group regarding
(formal) reviews.
If a reviewer has an issue (with something in what's being reviewed),
their issue will only
be accepted if they provide a solution.

As to the prevalence of terms from BFO, the book 'Building Ontologies
with BFO provides
a very nice list of terms and definitions (for which we received
permission to use).

I do agree that it would not be the most enlightening source for a
(generic) enterprise
customer. But enterprise architects may gain something from it, they
have to deal
with problems some of which are similar to those in ontology.

Todd

[Pat Hayes]

On Jul 6, 2017, at 4:45 AM, Matthew West <dr.matt...@gmail.com> wrote:

Dear Rob,

 

 

Replies in-line below...

On Wednesday, July 5, 2017 at 8:10:13 PM UTC+1, John F Sowa wrote:

Dear Matthew and Henson, 
> [MW>] There just is not one TLO everyone will agree to. However, 
> I think standardising a few so that they are comparable is a good 
> step towards overcoming the differences among them. 

The current proposal does not standardise a few, but has always aimed at only one. 

[MW>] That is not true. The standard is open ended as to the number of TLOs that can be standardised or claim conformance.

 If multiple TLOs are permitted, presumably not all agreeing on hogh-level categories, what constitutes ‘conformance’? What criteria are there for ‘conformance' of TLOs? 

Pat

No other tlo is participating in the design of the criteria in this proposal or in general.

[MW>] Well I am, and I have at least one TLO I might seek to either claim conformance for or standardise here.

 

Regards

 

Matthew West                            

Information  Junction

Mobile: +44 750 3385279

Skype: dr.matthew.west

matthe...@informationjunction.co.uk

http://www.informationjunction.co.uk/

https://www.matthew-west.org.uk/

This email originates from Information Junction Ltd. Registered in England and Wales No. 6632177. 

Registered office: 28, Connemara Crescent, Whiteley, Fareham, Hampshire, PO15 7BE.

 

 

[Matthew West]

Dear Pat,

 

On Jul 6, 2017, at 4:45 AM, Matthew West <dr.matt...@gmail.com> wrote:

 

Dear Rob,

 

 

Replies in-line below...

On Wednesday, July 5, 2017 at 8:10:13 PM UTC+1, John F Sowa wrote:

Dear Matthew and Henson, 
> [MW>] There just is not one TLO everyone will agree to. However, 
> I think standardising a few so that they are comparable is a good 
> step towards overcoming the differences among them. 

The current proposal does not standardise a few, but has always aimed at only one. 

[MW>] That is not true. The standard is open ended as to the number of TLOs that can be standardised or claim conformance.

 

 If multiple TLOs are permitted, presumably not all agreeing on hogh-level categories, what constitutes ‘conformance’? What criteria are there for ‘conformance' of TLOs? 

[MW>] That is being worked on at the moment, but it is mostly about a standard of documentation including definitions and documentation of axioms, plus an OWL and CL representation. Nothing unreasonable.

It’s worth noting that a TLO is smaller than most so called upper ontologies, really just the philosophical bit that most punters would probably wonder why it was there at all. This means that many Upper Ontology authors would need to extract the TLO from their upper ontology. Again, not a huge task.

I don’t recommend reading the current version available. It’ll probably irritate you more than anything. I’m hoping the next version will be worth reviewing, but will need review and revision before we have something fit for CD ballot. I’d happily bring any comments of yours to the table.

 

Matthew

[Patrick Cassidy]

The option John Sowa left out of his list is the one that I think has the only serious chance of promoting true general interoperability:

Develop a "upper" ("foundation") ontology that has all of the identifiable semantic primitives, such that those primitives are sufficient to specify logically the elements of all of the ontologies that it interoperates. Logically incompatible systems will exists as separate theories. That should rarely or never be necessary if the primitives are properly chosen, and local ontologies refrain from asserting that commonly used concepts (3D/4D) do not exist.

As new ontologies are linked together, new primitives may need to be added. **BUT** that will not break any of the interoperations of previously linked ontologies, since they do not need those new primitives.

One virtue of this approach is that any group that is independently developing its own ontology-based application need only have one member that is bilingual in the terms and meanings of their local ontology and the common foundation ontology.

The problem is that proving the interoperability of multiple ontology-based applications will require a program of substantial size (i.e. cost). Other than that, it should be straightforward.

PatCaassidy

>-----Original Message-----
>From: ontolo...@googlegroups.com [mailto:ontolog-
>fo...@googlegroups.com] On Behalf Of John F Sowa
>Sent: Tuesday, July 04, 2017 11:36 PM
>To: ontolo...@googlegroups.com

>Subject: Re: [ontolog-forum] Proposed ISO standard for ontology
>

[John F Sowa]

Todd, Pat H, David P, and Chris P,

Todd

> From a practical perspective l'll be pushing for consensus
> on common terminology (i.e., see the IAOA terminology list

> http://iaoaedu.cs.uct.ac.za/pmwiki.php?n=IAOAEdu.TermList .

That goal is not bad. But those of us who have looked at that list
have already reached a consensus: Kill it before it multiplies.

Pat

> this is a disaster. Both the selection of terminology to standardize

> and the definitions offered are completely idiosyncratic...

> To standardize on these would be simply to create another useless
> silo which will be ignored by some and enthusiastically treated
> as a gospel by others.

Chris
> I agree with David and Pat - this really needs quite a lot of work
> to make it usable.

David
> I agree 100% with Pat wrt the terms list... a few there are mostly
> really terrible (i.e. they would make the current situation worse)...
> the definition of “Ontology”... will do nothing but add confusion:

>
> Ontology: An ONTOLOGY is a representational artifact, comprising
> a taxonomy as proper part, whose representational units are intended
> to designate some combination of universals, defined classes, and
> certain relations between them

That is the worst definition of ontology I have ever seen. If you type
"define ontology" to Google, the first thing you get is much better:
> 1. the branch of metaphysics dealing with the nature of being.
>
> 2. a set of concepts and categories in a subject area or domain that
> shows their properties and the relations between them.
>
> Example: "what's new about our ontology is that it is created
> automatically from large datasets"

The Merriam-Webster definition, with one correction, would also be good:
> 1. a branch of metaphysics concerned with the nature and relations of
> being. Ontology deals with abstract entities.
>
> 2. a particular theory about the nature of being or the kinds of things
> that have existence.

The obvious correction is to say "both physical and abstract entities"
instead of just "abstract entities".

However, most dictionaries protect their copyrights. If we copy too
many definitions from them, IOAO could be sued. Fortunately, the
_Century Dictionary_ is out of copyright. And not by accident, the
definition of 'ontology' was written by Charles Sanders Peirce:
> The theory of being; that branch of metaphysics which investigates
> the nature of being and of the essence of things, both substances
> and accidents.

For Peirce's full definition and quotations by Watts, Hegel, and
Hamilton, see the attached file, ontology.jpg.

For other words, see http://www.global-language.com/century/ and ask
for the jpg. That returns the full page that includes the definition.

Recommendation: For each term in the IOAO list, cite several
definitions and their sources. To provide a broader range of
options, include the version from the Century Dictionary and
any others that may be available.

Then for each term, provide a comment area for members of IOAO or
Ontolog to state their suggestions, preferences, and revisions.
A few rounds of revisions, commentary, and voting should produce
some good definitions.

I'll volunteer to do the following:

1. Produce a list of definitions that can be posted alongside the
current crop that is now polluting the IOAO list.

2. I'll start with a glossary developed in 1997 by the NCITS T2
ontology working group: http://www.jfsowa.com/ontology/gloss.htm
That was a good group. See http://www-ksl.stanford.edu/onto-std/

3. And I'll also add any definitions from the _Century Dictionary_
to that list. Like the definition for 'ontology', many of them
were written by Peirce.

4. Then it would be useful if somebody would add a "blog" option so
that members of Ontolog and IAOA could make suggestions about which
definitions to extend/merge/revise/delete.

John

[John F Sowa]

On 7/7/2017 2:00 PM, Patrick Cassidy wrote:
> The option John Sowa left out of his list is the one that I think has
> the only serious chance of promoting true general interoperability:

We have been discussing that option for over 30 years, and my hopes
for finding any such primitives has decreased by about 3.3% per year.

> Develop an "upper" ("foundation") ontology that has all of the

> identifiable semantic primitives, such that those primitives are
> sufficient to specify logically the elements of all of the ontologies
> that it interoperates.

Language-based (lexical) resources, such as WordNet, are useful for NLP.
Longman's 2000 defining words are not primitive, they are not precisely
defined, and nobody has found them useful for an ontology. Those
resources have been used for aligning the mid levels of different
ontologies, but not for defining the top levels.

> The problem is that proving the interoperability of multiple
> ontology-based applications will require a program of substantial
> size (i.e. cost). Other than that, it should be straightforward.

Many people have devoted a lifetime of research to such projects
and led collaborations with fairly large groups of students and
colleagues. Igor Mel'chuk started in Russia in the 1950s and
continued in Canada since 1977. He developed Meaning-Text theory
supported with many volumes of lexical analysis. His work has
been useful for NLP and for machine translation -- but not ontology.

Anna Wierzbicka published her first set of primitives (Lingua Mentalis)
in 1972. She continued to publish books and articles on the issues
for over 40 years. Her analyses are impressive. But no one has
been able to use her primitives in a TLO.

Do you remember the Japanese 5th Generation Project? The Electronic
Dictionary Research (EDR) component cost over $100 million. It included
a dictionary of about 400,000 concepts with translations to both
Japanese and English. See http://www.wtec.org/loyola/kb/c5_s2.htm

Since CSLI at Stanford consulted on the project, they got a free copy
of the results. I spoke to people at CSLI, and nobody had found any
use for it. They did not include it in any of their R & D projects.

These projects produced useful results in linguistics, NLP, and MT.
But none of them discovered primitives that were suitable for TLOs.

John

[Todd Schneider]

In the context of information systems and knowledge representation my
understanding is that an ontology is an engineered artifact created to meet
some requirements and intended usage.

Most, if not all, engineering disciplines come to some consensus on the
terms
used (for the practice of their discipline) and their (expected)
interpretations.

Though the current field of ontology engineering has a range of
understandings
and interpretations of the terms used, a lack of some consensus bodes
poorly
for a larger acceptance and use.

Todd

[Edward Barkmeyer]

Todd,

Regrettably, the software engineering industry, and information technology in general, has a 60-year history of failure to develop a consensus for the terms used in its practice. Indeed, they prefer to re-invent old practices under new terms, so as to avoid creating much of an engineering discipline at all.

The AI world may be perhaps forgiven for their continuing inconsistencies in the use of the term 'ontology', because there is still a conflict between philosophical attitudes and knowledge engineering attitudes in the AI community itself. And it is further compounded by the phenomenon all researchers love and hate -- it relates to availability of funding.

-Ed

[Chris Partridge]

Or a lack of consensus is an indication of a that the field has not matured and stabilised - and a random stipulated agreement of the terms would be a poor substitute for maturing.

Also look at (e.g.) Nancy Cartwright's How the laws of physics lie to see how the idea of a simple consensus is illusory.

--
All contributions to this forum by its members are made under an open content license, open publication license, open source or free software license. Unless otherwise specified, all Ontolog Forum content shall be subject to the Creative Commons CC-BY-SA 4.0 License or its successors.
--- You received this message because you are subscribed to the Google Groups "ontolog-forum" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-forum+unsubscribe@googlegroups.com.

[David Price]

Hi Todd and Ed,

Following on Ed’s point, I would argue that almost no ontology in use today was “engineered” - certainly not in the same way that airplanes are engineered. If you search for “defined engineered” two items show up that are quite telling:

1. design and build (a machine or structure).
"the men who engineered the tunnel"

2. skilfully arrange for (something) to occur

Airplanes are created using definition 1. Ontologies, like data models of any other kind or in fact any other kind of software artefact, are created using definition 2 (i.e. developers skilfully arrange the Java code to read the input and produce the output).

Seems to me there are two reasons for this:

- engineered artefacts live in the real world and are subject to the laws of physics, chemistry, etc. Software lives in a virtual world based on very few, very low level “laws”.

- software companies have successfully lobbied to remove any liability for the failure of software, and so “engineering” has never developed in the discipline

Todd - to me in 2017 “Ontology" = "a data model written using a logic language” because that reflects the actual industrial practice. 10 years ago I’d have called that a “Strong Ontology” because many people tried to claim any data model was an ontology as described in

[1] The Ontology Spectrum & Semantic Models, Dr. Leo Orbst, MITRE, 2006

which we referenced in the ISO TC184 SC4 (ISO 10303 STEP and ISO 15926) project to set up something we called Industrial Data Integrated Ontologies and Models (IDIOM) [2] about 10 years ago.

There is very little philosophy in most enterprise ontologies so the definition, for example, from BFO has no wide industry buy-in whatsoever (and in fact cannot because it is so specific to BFO). Now I’d say “Strong Ontology” = “an Ontology that follows a prescribed set of metaphysical decisions and modelling practices” or something like that. Then, BFO Ontology is a “Strong Ontology that …”, etc. so perhaps a little progress has been made in the past 10 years.

Cheers,
David

[1] https://www.google.co.uk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0ahUKEwi2s8ee8IDVAhVL6xQKHYfsC0sQFggpMAA&url=http%3A%2F%2Fontolog.cim3.net%2Ffile%2Fresource%2Fpresentation%2FLeoObrst_20060112%2FOntologySpectrumSemanticModels--LeoObrst_20060112.ppt&usg=AFQjCNGaraXFUg3-HQUQd-1f3-kPUwm9CA

[2] https://www.researchgate.net/publication/269401880_ISO_TC_184SC4_N2614_Future_SC4_Architecture_PWI_Report_and_technical_discussion

UK +44 7788 561308
US +1 336 283 0606

[Patrick Cassidy]

None of the projects mentioned by John actually attempted to build a primitives-based foundation ontology and demonstrate that it can indeed enable interoperability among multiple independently developed ontology-based systems (or databases). I am familiar with much of the work John cited, but may have missed some critical part. If anyone can specifically refer to a project that attempted to use a primitives-based ontology to specify the meanings in other ontologies and thereby enable interoperability among them, I would appreciate a very specific internet-accessible reference. Just pointing to some project is useless - I have wasted a great amount of time trying to track down "relevant" information suggested by people, only invariably to find it utterly irrelevant.


The CYC baseCV is close to that notion, but not actually used in that way.

PatC

>-----Original Message-----
>From: ontolo...@googlegroups.com [mailto:ontolog-
>fo...@googlegroups.com] On Behalf Of John F Sowa
>Sent: Saturday, July 08, 2017 11:27 PM
>To: ontolo...@googlegroups.com
>Subject: Re: [ontolog-forum] Proposed ISO standard for ontology
>

[ontos Rob]

[MW>] Part 1 is trying to do one simple thing - set some standards for a TLO. Obviously there is much more to do, but that will need to wait for later standards.

This seems unlikely given that both the original draft and current one are bfo-focused. And that the content of the proposal is based on bfo and obo methods and concepts. And that it was not designed with input from various TLO developers or members of the international ontology community. Hence it is unlikely to truly for the community, i.e. for all tlo's.
This is why such a proposal--one truly intended to be for all--should involve *from the start* multiple tlo developers and views. Certainly with your good input and efforts it may be more fair, but given the context (origination, current status, lack of other tlo developers as editors, etc.) it's unlikely to be for the community.

[ontos Rob]

Some VERY important comments by experienced persons in the field are worth re-iterating (from this and the branched 'Glossary' thread):

"a lack of consensus is an indication of a that the field has not matured and stabilised - and a random stipulated agreement of the terms would be a poor substitute for maturing."

[Chris Partridge]

Precisely. Lack of consensus, the very discussion that is ongoing, and the below quotations, indicates the field needs to mature further. And there's nothing wrong with that!
This makes the current iso proposal premature, and unsuitable by design, particularly since it is based on the method and concepts of a single tlo and therefore does not represent the interests of community at large. Again, ISO is no place for a field to go through maturing or growing pains, particularly given the context of the particular proposal.

"inconsistencies in the use of the term 'ontology', because there is still a conflict between philosophical attitudes and knowledge engineering attitudes in the AI community itself..."
[Edward Barkmeyer]

Precisely. Same comment as above.

"Though the current field of ontology engineering has a range of understandings
and interpretations of the terms used,  a lack of some consensus bodes poorly

for a larger acceptance and use." [Todd Schneider]

Precisely. Same as above.

And from this thread:
"The current study and practice of ontology is multi-disciplinary. Each discipline and sub-discipline brings their own understanding and definitions." [Todd Schneider]

Precisely. Which is one reason why such a proposal should have a varied set of tlo developers as editors from the beginning (unlike the current proposal) which would better indicate that it is truly for all tlos. Arguably, no single tlo developer should have lead position on a proposal that supposedly is for developing standard for all tlo's. The current proposal did *not* involve input and consultation from various members of the international ontology community, so how could it be for all tlo's. Given that all tlo developers are not members of iso and therefore do not have editorial influence, it seems unlikely that this would change.

[Todd Schneider]

Chris,

I can't disagree that "...a lack of consensus is an indication of a that the field has not matured ...".

However, to facilitate its maturation I do think a list of common terms and their (potentially)
multiple definitions will help in the process.

The field on ontology engineering can only mature as common understanding and practices
are promulgated. This is happening at universities and should be accessible to all at some
level.

Again, there was no expectation that the IAOA term list would be the 'last word'. As part of
the IAOA Education committee the intent was to help the larger community, even if they don't
agree with the definitions (in which case we would like to hear why). Understanding the
differences and assumptions or contexts for those differences would also foster the maturation
process.

Thoughts.

Todd

P.S. Is there a URL to Nancy Cartwright article?

To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.

[Chris Partridge]

Hi Todd,

My underlying concern is premature standardisation. 

I am very suspicious of standards-led maturity, it seems to me to put the cart before the horse in a very pernicious way - in that it moves to close down the maturing process - and one is stuck with a dud.

The Cartwright book is http://www.oxfordscholarship.com/view/10.1093/0198247044.001.0001/acprof-9780198247043.

Another good reference (which I think I have posted to the list before) is The Function of Measurement in Modern Physical Science by Thomas S. Kulm ~ Isis, Vol. 52, N0. 2 (Jun., 1961), 161-193.

This suggests the rigour or measurement emerges from understanding, rather than the other way around.

Chris

 

[rrov...@buffalo.edu]

Completely agree with Chris. ISO is not place for a field to mature or go through growing pains. The potential consequences, likely negative, will be felt by not just the field, but users, data, etc.

[Todd Schneider]

Ed,

Yes. I've seen this terminology problem in software all too often.
One cause I could hypothesize is a lack of knowledge (of what's
been done in other fields).

I would think that a list of terms, their (possibly multiple) definitions,
and the context of/for each definition would help is promulgating
(some) knowledge.

Todd

[Todd Schneider]

Chris,

Too your point about the lack of engineering of ontologies today,
unfortunately
there is ample evidence to support your point. But consider the
following definitions
one of which you referred to (from the OSX Dictionary):

Engineer: (Verb) design and build (a machine or structure)
Artifact/Artefact: (noun) an object made by a human being

Certainly an ontology is 'created' by humans or software devised by humans
and I think what is 'created' is an artifact of that process.

Now whether the process of 'creating' an ontology adheres to any principles
is another matter. And I think you captured some of this with "... "a

data model
written using a logic language” because that reflects the actual
industrial practice."

I'm not sure what was intended by "... very little philosophy in most
enterprise ontologies ...".
Was it that the creators of ontologies for commercial enterprises are
unfamiliar with
the philosophical aspects of (current) ontology? Or something else? I
know a few people
that develop enterprise ontologies that are aware of the philosophical
aspects (you probably
know that also). But in engineering an ontology pragmatic requirements
may take priority.

I think the book on developing ontologies with BFO is an excellent step
in the direction
of providing engineering principles for ontology development. Note that
this book has
a glossary, of course many the definitions presume the assumptions of
BFO. So have a
glossary of terms that can provide 'alternate' definitions, with context
for the definition,
would aid in developing the field (of ontology and ontological
engineering). I should make
note of two initiatives underway to continue this trend (Barry Smith is
involved in both).

0) ST4SE - Semantic Technologies for Systems Engineering
1) IOF - Industry Ontology Foundry

Each of these initiatives are still in an 'incubator' phase so you won't
find much, if anything,
available on the web.

Todd

[Todd Schneider]

Chris,

Thank you for the references.

As to 'standardization'. The intent of the IAOA's term list/glossary is for
standardization. But for knowledge(?) dissemination.

Ontology (in its current state - information systems & knowledge representation)
is a multi/inter-disciplinary field. Terms that occur in the different fields may have
variations in their definitions. Understanding the variations, or at the least being
aware of them and their differences, I think is useful.

An occurring problem I see (and not just in ontology development) is understanding
the scope of a problem, or in the case of ontologies, the (potential) scope of interpretation.

Todd

To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.

[Todd Schneider]

Robert,

I've made no mention of ISO and did not intended to suggest the
goal of the IAOA's term list/glossary is standardization. 
The intent is to foster consensus (which I think is not equivalent to
standardization) and for education (or at least amusement:).

Todd

On 7/12/17 6:41 AM, rrov...@buffalo.edu wrote:

Completely agree with Chris. ISO is not place for a field to mature or go through growing pains. The potential consequences, likely negative, will be felt by not just the field, but users, data, etc.

--
All contributions to this forum by its members are made under an open content license, open publication license, open source or free software license. Unless otherwise specified, all Ontolog Forum content shall be subject to the Creative Commons CC-BY-SA 4.0 License or its successors.
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.

[Hans Polzer]

Todd,

 

I think you intended to say that the intent of the IAOA’s term list/glossary is not for standardization, but rather for knowledge dissemination.

 

Hans

 

From: ontolo...@googlegroups.com [mailto:ontolo...@googlegroups.com] On Behalf Of Todd Schneider

Sent: Wednesday, July 12, 2017 2:01 PM
To: ontolo...@googlegroups.com

[Todd Schneider]

Hans,

Most definitely. I meant to write,

As to 'standardization'. The intent of the IAOA's term list/glossary is

NOT for standardization. But for knowledge(?) dissemination.

Thank you for catching that.

Todd