Hi Pat, (01)
> -----Original Message-----
> From: Pat Hayes [mailto:phayes@xxxxxxx]
> Sent: 20 December 2012 09:35
> To: Chris Partridge
> Cc: [ontolog-forum]
> Subject: Re: [ontolog-forum] Fwd: Ontologies and individuals
>
>
> On Dec 19, 2012, at 5:28 PM, Chris Partridge wrote:
>
> > Hi John,
> >
> > That was not my initial reading of Pat's response, and after
> > re-reading I still cannot not read it in this way.
> >
> > As I understand it, Pat was saying that the philosophical terms, such
> > as the ones in your foundation (see
> > http://www.jfsowa.com/ontology/toplevel.htm - note the use of the terms
> 'abstract' and 'physical'), are:
> > "indeed obscure, hard to decide, and rest upon very fragile intuitions.
...
> > they are not foundational in the sense that anything important rests
> > upon them or depends on getting them right. They are, in fact, almost
> > completely irrelevant to any matter of concern to practical ontology
> > design, and can (and should) be ignored, unless one has far too much
> > time to waste. It is not an architectural decision at all: more like a
> > decision about what color smoke to blow up the chimney."
>
> I was intending to refer in particular to the individual/non-individual
> distinction which has been the topic of this thread, but OK, I tend to be
> dismissive of many high-level distinctions, so let us proceed on that
basis. (Of
> all of them, the abstract/physical is probably the most intuitively
secure, but
> even there one rapidly gets into difficulties in particular cases.) (02)
I'm intrigued by your comment "the abstract/physical is probably the most
intuitively secure ". It is a modern intuition, in philosophy at least "If
the abstract/physical distinction is intuitive, it is a modern intuition"
and it is notoriously difficult to define "... there is no standard account
of how it should be drawn. " both from
http://plato.stanford.edu/entries/abstract-objects/ .
>From my reading, I would say that the general/specific
(universal/particular) in the more ingrained (after all it is coded into
natural language) and ancient (hence my earlier comment about Aristotle)
intuition - but that there have been a large variety of attempts (in
philosophy) to explain what it actually is. Whereas, the problem with the
abstract/concrete distinction is getting a definition for it.
I'm also surprised to find you on the abstract realist side of the debate;
I'd have assumed that if you were going to take a position it would be with
Quine, Field, etc.
Furthermore, I'm a little surprised at your taking intuition as any kind of
guide. Given how easy it is to train and fool, I'd be wary about raw
intuition without some backing. (03)
>
> > And these comments apply equally to similar work, such as the work of
> > Matthew on ISO 15926, Nicola and Chris on Ontoclean and Barry on BFO,
> > which are also based on philosophical ideas.
> >
> > Maybe Pat can clarify whether I have misread this.
>
> Yes and no. I apparently misunderstood your use of "foundation" to mean
> "foundational", as you correctly diagnosed (below) and I was reacting in
part
> to that claim (which apparently you did not intend...although from your
> reaction here, maybe you did in fact intend? I am no longer sure... but
> anyway...).
I was just trying to make a point about how Matthew was setting up his
foundations - I think that they can be done structured in a better way.
I do however, think that we can learn from philosophy and that it is a good
input into the foundations one might choose to build. (04)
I am of the view, which you know I have often expressed and
> which I know I share with others including Doug Lenat and John Sowa (some
> of the time, at any rate), the view that the very high levels of most
ontologies (05)
Not all??? (06)
> make distinctions which are not of any real importance to the lower
levels, in
> the exact sense that those lower levels could be re-written so as to
conform
> to a different set of high-level distinctions without materially affecting
any
> entailments they have about any matters of substance. Whether one chooses
> to divide up the world initially into (say) individuals versus
non-individuals, or
> physical versus abstract, or occurrent versus continuant, and so on (and
on
> and on), is, contrary to what one would expect from the amount of
attention
> such decisions get, of really rather little importance. Whatever one
decides,
> there will be cases that do not fit comfortably into the taxonomy one has
> devized, and cause difficulties which have quite recognizable symptoms,
> including: essentially the same content having to be re-stated in two
different
> ways; various odd corner or exception cases being included in the class
> hierarchies; entire sections of reality being described in ways that seem
> strange, idiosyncratic or just plain wrong to users who have not been
trained
> to think in the framework provided. (In extreme cases, the only way to
decide
> how to use the ontology is to ask one of the inventors, who has a 'guru'
> status in the community of users. Barry is the guru of BFO, for example,
as
> can be seen from reading the email logs, where he is frequently asked to
rule
> on whether some entity is a continuant or not, as nobody else seems able
to
> decide.) (07)
It is true that if one aims for a consistent organisation over a large
amount of data, one is faced with situations where the local fit can be
difficult. However, this seems to be in the nature of these tasks, however
one tried to organise it, that there is a trade-off between getting a wider
consistency and current local practices.
My experience is that the better the foundation, while there may be some
effort required to find a conformant formulation (or indeed any logical
formulation), the less need there is for a workaround. To quote Plato, the
classifications seem to cut nature at its joints - that is not to suggest
that there are different ways to cut it up. (08)
>
> Now, this may be the best we can do, in practical terms. And I concede the
> argument that in order to make progress, one has to adopt *some*
> framework and stick to it, so that interoperability can be achieved in a
large,
> heterogenous community of users. And maybe the emergence of 'go-to' guys
> to clear up confusions and puzzles is inevitable and should be encouraged
> rather than deplored or mocked. But let us not confuse this kind of
> managerial pragmatics with philosophy, and believe that any of these
> basically arbitrary top-level taxonomies is somehow *true* or is a genuine
> abstraction from *reality*. (09)
I think 'arbitrary' is too strong. I would agree that one's choice should be
driven by engineering pragmatics. I think that it is not too farfetched to
think that the reason the top levels work is that they reflect some features
of reality. If they are completely arbitrary, then why do they work?
However, I would agree that whether this is real *reality* is not relevant -
in the same way many engineers look at the science upon which their
engineering is based - the way a civil engineer uses Newtonian physics and
ignores relativity and quantum mechanics. I cannot see any real harm in
people wanting to believe it is really real - except that they then feel a
moral duty to try and undermine other top frameworks. Or in your case,
undermine ALL top frameworks :-). (010)
If you have made individual/non-individual into a
> basic distinction, and want your users to respect it, then fine. But I
wouild ask
> for three things. (1) recognize, and admit, that this distinction is local
to your
> perspective and not some kind of absolute, or justified by philosophical
> necessity, or given from the ontological gods (such as Aristotle); (011)
I would agree that if anyone was making an argument from philosophical
authority then this is 'wrong'. The reason being not so much that the
philosophy might be wrong, but that it may not be the appropriate tool for
the task at hand. However, I think it would be perverse to say that if one
found exactly the tool one wanted in philosophy that one should not use it.
Or, when examining a topic that has been researched by philosophy that one
should on no account use that research. (012)
(2) explain
> what you mean by it as clearly and non-technically as possible, especially
if it
> differs from what others mean by this language (which it almost certainly
will)
> and (3) explain, or at least justify, WHY you are making this distinction,
in
> ontological terms. What difference does it make to classify something this
> way rather than that? What follows, or is prevented from following, by
such
> decisions? Why do you feel a need to make this distinction at all? And so
on. (013)
WRT (2) and (3): are these not exactly what I was suggesting when I said:
>>> So, to make sense of this I think it helps to expose the
>>> architectural choices one had made - and as the discussion shows (a)
>>> we often do not do this and (b) there are a myriad of choices to choose
from.
I was suggesting a methodical way of doing it. One which recognises that the
choices one makes often influence one another. One which you dismissed in
highly emotional terms as totally irrelevant - as I recall. (014)
>
> In a quieter moment I will concede that top-level distinctions may be of
some
> limited utility. But I certainly believe that they receive far more time
and
> attention than is justified by their importance. As soon as one gets into
the
> details of virtually any topic, these top-level distinctions fade into
> insignificance compared to other matters which need attention, at best,
and
> they often actively impede progress, at worst, because apparently clear
> decisions about the nature of things often become far less certain and far
less
> clear when one is in the thickets of the details. (015)
My experience in this is quite different from yours. I have regularly
attended meetings (and also heard about meetings) where it takes a group a
week to agree a single term. When a top ontology has been introduced, the
turnaround has reduced significantly. The top ontology gives a framework for
focussing the discussion. Maybe we are dealing with different audiences.
Maybe, in a sense, we are agreeing. I'm suggesting that having the top
ontology in place means that discussion of these topics is no longer
necessary, that one can focus on the lower level details. Of course, there
are then issues about getting a top ontology that facilitates this. I also
think difference in the domain is relevant. My interest is in domains that
are already automated, where there is significant data and much of it is
processed by machines. In the medical domain that Barry looks at, much of
the data is unstructured and not processed automatically by machines. These
may well require different top ontologies. (016)
Is a time-interval abstract or
> concrete? (What about a physical unit or a measuring scale?) Is a work of
> literature a continuant or an occurrent? Not only are such questions hard
to
> answer, they are also not very interesting or important. They are
questions
> about the classification system itself, not about the things being
classified. We
> all have intuitions about time-intervals and literary works which are
robust
> enough to be axiomatized and utilized without ever asking these
classification
> questions, and which do not provide answers to them. (017)
In my experience, most of what you say above is not true. And what have
intuitions got to do with this? (018)
But in frameworks
> (like OBO) which *require* one to make such categorizations in order to
even
> begin (since one is obliged to use different formal techniques in each
case),
> these questions pose constant problems, problems which are completely
> artificial.
>
> To return to "individual". If this is taken to mean "not having instances"
(or
> maybe, "of a kind that inherently does not have instances", to rule out
the
> empty set) then my problem with it is that the actual logic of CL makes
> *everything* into a non-individual, because CL allows any name to be used
in
> a relation or predicate position. (019)
Isn't CL just a tool for implementing ontologies. I see no reason why one
cannot implement an ontology that recognises the individual-non-individual
distinction in CL - in fact, I have been part of a team that considered
doing just this.
Why would anyone want to build a tool that stopped people implanting their
ontologies in it? (020)
And IMO this is the right way to approach
> this issue: it is not an ontological matter at all, but a *logical*
matter. The
> actual language of logic is based upon the idea of a predication by one
name
> of a sequence of other names: an atomic sentence. So to claim that there
is a
> basic distinction between the predicators and the predicatees is to say
that
> the relation and (ahem) 'individual' names should be segregated in one's
> logic. But there are important, down-to-earth reasons for using an
> unsegregated logical language, reasons that far transcend any
philosophical
> scruples about the nature of non-individual-hood. Basically, one gets the
> expressive advantages of a higher-order language within a purely
first-order
> logic. Such a gain is not to be lightly dismissed (for example, it means
that one
> can replace all the OBO duplication of axioms for the occurrent and
> continuant cases by a single scheme which covers both cases, making that
> distinction logically vacuous); but more fundamentally, the possiblity of
using
> an unsegregated first-order language seems to render this whole discussion
> moot: for what is the point of deciding on a classification which is
rendered
> vacuous *as a matter of logical necessity* when one uses a modern logical
> notation? Aren't there more important things to worry about? Why not just
> say, you can think of anything as an individual (when you wish to make
> assertions about it) and also as a predicate or relation (when you wish to
use
> it make assertions about other things), and this choice of of no
ontological
> importance whatsoever. Because if you write your axioms in CLIF, it isn't. (021)
Aren't these all arguments about designing a neutral ontological language -
not about which top ontology, if any, to choose. (022)
>
> Pat
>
> >
> > BTW I have no objection to someone taking this position, it makes
> > sense to explore different options. But given the substantial
> > contributions made (by you and others) it seems to me that one should
> > make a rational rather than tabloid argument for it.
> >
> > As I understand it, Pat was proposing the following as an alternative
> > to the kinds of top ontology foundations referred to above.
> > "Here is one way to decide the matter: anything that can be described
> > or referred to is, ipso facto, an individual. And of things that
> > cannot be described or named, we must be silent."
> > I assume he was suggesting that the philosophical terms in, for
> > example, your top ontology would fail this test. Presumably, if they
> > did not then they would pass the Hayes test.
> >
> > None of my comments were to do with words. I certainly agree that
> > there are problems with the senses of words used in this community
> > (as I hope I made clear in my responses), but I am not offering a
> > solution; I suspect there is not an easy one.
> >
> > Also, I hope I do not need to make clear that I admire a lot of Pat's
> > work - not only have I said that to him personally a number of times,
> > but it should be clear from previous email exchanges. That I assume is
taken
> for granted.
> >
> > And just to clarify Pat's response was nothing to do with my original
point:
> > Matthew had described a series of choices and I said relative to this
> > description: "If you have made these two choices then a mental
> > shorthand for the distinction might be that it rests upon a
> > concrete-abstract distinction
> > - but (architecturally) this is a much weaker and murkier distinction
> > to base things on, so not such a solid foundation." I was using
'foundation'
> > here in the sense of one choice underpinning another - not implying
> > that any choice was foundational.
> > Pat chose to change the topic to whether the choices Matthew was
> > describing were 'foundational'.
> >
> > Regards,
> > Chris
> >
> >
> >> -----Original Message-----
> >> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
> >> bounces@xxxxxxxxxxxxxxxx] On Behalf Of John F Sowa
> >> Sent: 19 December 2012 18:23
> >> To: ontolog-forum@xxxxxxxxxxxxxxxx
> >> Subject: Re: [ontolog-forum] Fwd: Ontologies and individuals
> >>
> >> On 12/19/2012 3:56 AM, Chris Partridge wrote:
> >>> I know you [Pat] have an irrational aversion to these topics.
> >>
> >> That remark is uncalled for. Pat has a very strong understanding of
> >> the
> > issues
> >> and requirements for a good design. But he has even less patience
> >> than I
> > have
> >> for endless quibbling over words.
> >>
> >>> My and, I'm guessing, Matthew's experience is that these kinds of
> >>> structures are a useful practical device when dealing with large
> > systems.
> >>
> >> Structure is *extremely* important. Pat and I were not complaining
> >> about
> > any
> >> proposed structures. We were complaining about endless debates about
> >> common English words that are overloaded with confusing and
> >> conflicting definitions.
> >>
> >> My solution is simple: whenever a word creates more confusion than
> >> enlightenment, delete it. If you can't state a precise definition in
> > terms of the
> >> logic used to specify the ontology, then the word is too vague and
> > misleading
> >> for ontology design and documentation.
> >>
> >>> If you [Pat] are suggesting that the terms we are using "cannot be
> >>> described or named" maybe you should re-read the literature.
> >>
> >> Pat not only knows the literature, he has been making important
> > contributions
> >> to it for over 40 years. The fact that this thread has been going
> >> around
> > in
> >> circles for dozens of notes without reaching a precise definition of
> >> the
> > word
> >> 'individual' implies that the word is more confusing than informative.
> >>
> >> Simple solution: If knowledgeable people in the field cannot agree
> >> on a definition for a word, then don't use it.
> >>
> >> By the way, I learned that lesson many years ago in a project at IBM.
> >> We
> > had
> >> endless arguments about the word 'name'.
> >> Our solution was to forbid any use of that word.
> >>
> >> When we got rid of the word, we quickly discovered that we all had
> >> nearly identical intuitions about the *structure* of the
> >> architecture. The only arguments were about one confusing word.
> >>
> >> John
> >>
> >>
> ________________________________________________________________
> _
> >> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> >> Config Subscr:
> >> http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> >> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> >> Shared Files: http://ontolog.cim3.net/file/ Community Wiki:
> >> http://ontolog.cim3.net/wiki/ To join: http://ontolog.cim3.net/cgi-
> >> bin/wiki.pl?WikiHomePage#nid1J
> >>
> >
> >
> >
> ________________________________________________________________
> _
> > Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> > Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> > Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> > Shared Files: http://ontolog.cim3.net/file/ Community Wiki:
> > http://ontolog.cim3.net/wiki/ To join:
> > http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> >
> >
>
> ------------------------------------------------------------
> IHMC (850)434 8903 or (650)494 3973
> 40 South Alcaniz St. (850)202 4416 office
> Pensacola (850)202 4440 fax
> FL 32502 (850)291 0667 mobile
> phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
>
>
> (023)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (024)
|