ICE and its concretizations
Bjoern Peters
It was my understanding that:
1) Every information content entity (ICE) is_concretized_as some (physical) quality. The ICE could be a weight measurement datum. The concretization is the ink pattern written down in a lab notebook, the magnetic pattern in a hard disk, and the 'neural pattern' in someones brain.
2) Each of these qualities 'inheres in' an independent continuant (the paper, magnetic storage, brain)
3) There is no 'inheres in' relation between an ICE and an independent continuant other than to the bearers of the concretized ICE qualities
I noted (as did MC during the call) that we have been treating 'plan' different from the above. We say that a plan is a concretization of a plan specification, and that a plan is a 'realizable entity, not a quality.
I think it is straightforward to make dealing with plans consistent with the above, and that it is desirable. For every plan that is realized, the bearer must be capable of storing information about that plan. It is possible to bear concretizations of plan specifications without the intend or capacity to realize them, like the piece of paper on which a plan specification is written.
Therefore i would propose to modify the definition of plan to be:
plan=def a realizable entity that inheres_in a bearer that has_quality a concretization of a plan specification which the bearer is committed to realize as a planned process.
--
Bjoern Peters
Assistant Member
La Jolla Institute for Allergy and Immunology
9420 Athena Circle
La Jolla, CA 92037, USA
Tel: 858/752-6914
Fax: 858/752-6987
http://www.liai.org/pages/faculty-peters
Pat Hayes
On Dec 8, 2009, at 1:06 PM, Bjoern Peters wrote:
> Based on some of the email conversations in the past days, and
> confirmed in todays call, we thought it is necessary to define
> properly some of the early workshop decisions for IAO which not
> everybody is aware of, and there seem to be some disagreements about.
>
> It was my understanding that:
>
> 1) Every information content entity (ICE) is_concretized_as some
> (physical) quality.
thing? I have been taking it that a concretization is something
existing in physical time and space, something that could be destroyed
by a physical process. Some of them (eg the Hollywood sign) might have
physical properties such as mass. Is this understanding mistaken?
> The ICE could be a weight measurement datum. The concretization is
> the ink pattern written down in a lab notebook, the magnetic pattern
> in a hard disk, and the 'neural pattern' in someones brain.
**pattern** distinct from the actual ink marks on a surface? If the
ink used was Indian ink, the marks contain carbon. Does the pattern
contain carbon?
I presume that, in some cases, two different ICEs might be concretized
in the same IC simultaneously (as in a coded message, for example).
Right?
>
> 2) Each of these qualities 'inheres in' an independent continuant
> (the paper, magnetic storage, brain)
particular piece of paper, rather than paper in general, right?
>
> 3) There is no 'inheres in' relation between an ICE and an
> independent continuant other than to the bearers of the concretized
> ICE qualities
Pat
>
> I noted (as did MC during the call) that we have been treating
> 'plan' different from the above. We say that a plan is a
> concretization of a plan specification, and that a plan is a
> 'realizable entity, not a quality.
>
> I think it is straightforward to make dealing with plans consistent
> with the above, and that it is desirable. For every plan that is
> realized, the bearer must be capable of storing information about
> that plan. It is possible to bear concretizations of plan
> specifications without the intend or capacity to realize them, like
> the piece of paper on which a plan specification is written.
>
> Therefore i would propose to modify the definition of plan to be:
>
> plan=def a realizable entity that inheres_in a bearer that
> has_quality a concretization of a plan specification which the
> bearer is committed to realize as a planned process.
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> --
> Bjoern Peters
> Assistant Member
> La Jolla Institute for Allergy and Immunology
> 9420 Athena Circle
> La Jolla, CA 92037, USA
> Tel: 858/752-6914
> Fax: 858/752-6987
> http://www.liai.org/pages/faculty-peters
>
> informatio...@googlegroups.com
> To change settings, visit
> http://groups.google.com/group/information-ontology
>
------------------------------------------------------------
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
Bjoern Peters
----- "Pat Hayes" <pha...@ihmc.us> wrote:
> > 1) Every information content entity (ICE) is_concretized_as some
> > (physical) quality.
>
> Quality? Does that imply something other than the actual physical
> thing? I have been taking it that a concretization is something
> existing in physical time and space, something that could be destroyed
> by a physical process. Some of them (eg the Hollywood sign) might have
> physical properties such as mass. Is this understanding mistaken?
>
> > The ICE could be a weight measurement datum. The concretization is
> > the ink pattern written down in a lab notebook, the magnetic pattern
> > in a hard disk, and the 'neural pattern' in someones brain.
>
> All of these are 'patterns', which raises doubts in my mind. Is an ink
> **pattern** distinct from the actual ink marks on a surface? If the
> ink used was Indian ink, the marks contain carbon. Does the pattern
> contain carbon?
> I presume that, in some cases, two different ICEs might be concretized
>
> in the same IC simultaneously (as in a coded message, for example).
> Right?
> >
> > 2) Each of these qualities 'inheres in' an independent continuant
> > (the paper, magnetic storage, brain)
>
> OK. The continuant in which the ink quality inheres would be the
> particular piece of paper, rather than paper in general, right?
>
Pat Hayes
On Dec 8, 2009, at 3:29 PM, Bjoern Peters wrote:
> Hi Pat - replies below
>
> ----- "Pat Hayes" <pha...@ihmc.us> wrote:
>
>>> 1) Every information content entity (ICE) is_concretized_as some
>>> (physical) quality.
>>
>> Quality? Does that imply something other than the actual physical
>> thing? I have been taking it that a concretization is something
>> existing in physical time and space, something that could be
>> destroyed
>> by a physical process. Some of them (eg the Hollywood sign) might
>> have
>> physical properties such as mass. Is this understanding mistaken?
>
> 'quality' is referring to BFO:quality, which is tightly linked but
> not the same as a independent continuant (='material entity' or
> physcial entity for the discussion here). For example, a human being
> that has different mass at different time points is considered the
> same material entity with different mass qualities.
very far), such a quality would be a property, or perhaps a fluent, ie
a property varying with time, of the physical entity. Right? And a
property, or property-like thing, is apparently not itself a physical
thing. So, to answer my other question, a concretization which is a
*pattern* of ink marks does not itself contain small particles of
carbon, even though the marks themselves do. (Because a property, or
quality, is not itself the kind of thing that can possibly contain
particles of carbon.)
I confess to being more confused now than I was before, in spite of
your valiant efforts to explain.
This is, for me, a perfect illustration of the reason why ontology
should not be done by philosophers. I know what a piece of paper is,
and I know what an ink mark is. I can understand the idea of an ink
mark being a rendering of writing and hence of a message with content.
I can make sense of talk which refers to properties of things like
this. But I have absolutely no idea what continuants and qualities
are, nor do I believe that I should need to, in order to understand an
ontology of ICEs. (Evidence: I have been thinking about and working
with ICEs for some time now, apparently successfully, without ever
even being aware of the notions of continuant or quality.) I cannot
believe that any ontology or even informal exposition that requires me
to learn and accede to an alien set of philosophical abstractions can
possibly provide any kind of enlightenment about the nature of part of
reality.
I think it might be best if I simply remove myself from this group.
Almost every notion being discussed here is so alien to my way of
thinking that it seems clear that I will only be sand in the bearings
if I try to participate. For example, although I have been trying hard
not to let this stand in the way, I simply do not believe that
continuants exist, or are even possible. The very idea seems to me to
be incoherent. I am already reconciled to not being able to understand
or make use of anything which is rooted in BFO's strange, surreal,
view of the world, and it would obviously be counterproductive to
attempt to change it when this group has already decided to work
within it.
Good luck, y'all, in this mysterious wonderland of qualities and
abstract concretion and independent continuants. I will stick to
relationships between entities with spatiotemporal extents, where I
know what I am talking about.
Pat
Melanie Courtot
On 8-Dec-09, at 1:29 PM, Bjoern Peters wrote:
> Hi Pat - replies below
>
> ----- "Pat Hayes" <pha...@ihmc.us> wrote:
>
>>> 1) Every information content entity (ICE) is_concretized_as some
>>> (physical) quality.
>>
>> Quality? Does that imply something other than the actual physical
>> thing? I have been taking it that a concretization is something
>> existing in physical time and space, something that could be
>> destroyed
>> by a physical process. Some of them (eg the Hollywood sign) might
>> have
>> physical properties such as mass. Is this understanding mistaken?
>
> 'quality' is referring to BFO:quality, which is tightly linked but
> not the same as a independent continuant (='material entity' or
> physcial entity for the discussion here). For example, a human being
> that has different mass at different time points is considered the
> same material entity with different mass qualities.
of this quality fluctuates, but not that there are different mass
qualities.
The definition of bfo:quality says "A specifically dependent
continuant [snap:SpecificallyDependentContinuant] that is exhibited if
it inheres in an entity or entities at all (a categorical property)."
The definition of bfo:realizable entity says: "A specifically
dependent continuant [snap:SpecificallyDependentContinuant] that
inheres in continuant [snap:Continuant] entities and are not exhibited
in full at every time in which it inheres in an entity or group of
entities. The exhibition or actualization of a realizable entity is a
particular manifestation, functioning or process that occurs under
certain circumstances."
(In BFO, quality and realizable entity are disjoint.)
Based on this, I would say that every ICE is concretized as some
realizable entity. For example when concretizing a plan specification,
a specific neural pattern is involved in my brain. This pattern is not
exhibited in full at every time (I am currently living without the
plan of what I will do tomorrow). To execute the plan, I will perform
a specific process to that effect, which will realize this entity.
>
>>
>>> The ICE could be a weight measurement datum. The concretization is
>>> the ink pattern written down in a lab notebook, the magnetic pattern
>>> in a hard disk, and the 'neural pattern' in someones brain.
pattern that I have formed with respect to the plan specification is a
quality.
Melanie
TFL- BCCRC
675 West 10th Avenue
Vancouver, BC
V5Z 1L3, Canada
Bjoern Peters
Getting used to BFO has been weird for me as well, but it worked well so far (for me at least). I would agree though that it will be better if you re-join this conversation at a later time when there is a higher degree of maturity, so that we don't have to develop already complicated enough issues using different frameworks.
- Bjoern
>
> > Hi Pat - replies below
> >
> > ----- "Pat Hayes" <pha...@ihmc.us> wrote:
> >
> >>> 1) Every information content entity (ICE) is_concretized_as some
> >>> (physical) quality.
> >>
> >> Quality? Does that imply something other than the actual physical
> >> thing? I have been taking it that a concretization is something
> >> existing in physical time and space, something that could be
> >> destroyed
> >> by a physical process. Some of them (eg the Hollywood sign) might
>
> >> have
> >> physical properties such as mass. Is this understanding mistaken?
> >
> > 'quality' is referring to BFO:quality, which is tightly linked but
>
> > not the same as a independent continuant (='material entity' or
> > physcial entity for the discussion here). For example, a human being
>
> > that has different mass at different time points is considered the
>
> > same material entity with different mass qualities.
>
Bjoern Peters
realizable entity"
If I write a measurement datum on a piece of paper (which is a concretization of an ICE) what is the realizable entity?
Melanie Courtot
- the ICE is the measurement datum
- the realizable entity is "the specific neural pattern which inheres
in my brain and is exhibited during the process of me writing down on
the piece of paper this ICE"
- this realizable entity is realized during a writing process, which
outputs a material information bearer (piece of paper) that
has_quality information carrier (pattern of ink)
- the dependent continuant measurement datum inheres_in the piece of
paper that has_quality pattern of ink. I wouldn't say that the piece
of paper is a concretization of an ICE, I would say the piece of paper
bears the ICE.
I disagree when you say that a concretization of a plan specification
is a quality, in
>>>>> has_quality a concretization of a plan specification which the
>>>>> bearer is committed to realize as a planned process.
categorical property as per the BFO definition?
Maybe it should be that every ICE is concretized as some SDC?
Melanie
Bjoern Peters
> If I write a measurement datum on a piece of paper:
> - the ICE is the measurement datum
> - the realizable entity is "the specific neural pattern which inheres
> in my brain and is exhibited during the process of me writing down on
> the piece of paper this ICE"
> - this realizable entity is realized during a writing process, which
> outputs a material information bearer (piece of paper) that
> has_quality information carrier (pattern of ink)
> - the dependent continuant measurement datum inheres_in the piece of
> paper that has_quality pattern of ink. I wouldn't say that the piece
> of paper is a concretization of an ICE, I would say the piece of paper
> bears the ICE.
>
> I disagree when you say that a concretization of a plan specification
>
> is a quality, in
> >>>>> plan=def a realizable entity that inheres_in a bearer that
> >>>>> has_quality a concretization of a plan specification which the
> >>>>> bearer is committed to realize as a planned process.
>
>
> Could you describe what is the quality in this case and how it is a
> categorical property as per the BFO definition?
Melanie Courtot
In your original email, you were saying that a human being at
different points in time has different mass qualities. I would say
that a human being has one mass quality, which value fluctuates over
time, depending on the amount of matter constituting that human.
Similarly, I could say that a human being has a quality of being able
to form neural patterns (alternatively a brain has function creating
neural patterns), but I wouldn't say a specific neural pattern is a
quality of the human (or its brain).
Melanie
Bjoern Peters
I want to stick to written things on paper for now, as I don't know too much about neuron pattern formation:
One piece of paper. I write a sentence on the top. You write one on the bottom. How many 'ink pattern qualities' does the paper have? I would have thought two, but I think I can live with one. In that case, I would rephrase that the upper part of the paper has an ink pattern quality, and the lower part has another. So there can be two ICEs concretized on the same piece of paper, one on the upper part, one on the lower part.
I believe the same can be said for harddisk patterns and neural patterns (although for the latter we have little knowledge which part of the brain contains what knowledge).
Some other points:
- It should definitely be a brain function to from neural patterns, not a quality of the brain.
- In your original justification for why a concretization of an ICE is a realizable, you stated that it is realized in your process of writing it down. That would make a concretization of a plan specification is realizable in two completely different process: writing it down and executing the plan. I don't think that is accurate.
Jonathan Rees
In what forum is BFO being critically appraised? As BFO seems to be
shaping the whole foundry effort, including IAO and OBI, it would seem
wise to attempt to predict whether it's going to serve the endeavor
well, what its limits will turn out to be, and whether its reliability
or applicability might be increased (or threatened) by changes to it.
This is true even if a sunk cost argument convinces us all not to make
any changes to it.
Of course I mean an analysis with practical utility, not philosophical
considerations, as the test (admitting of course that the latter are
intended to predict the former).
If an appraisal gives BFO high marks, then it should have the benefit
of bringing skeptics like Pat and me into the fold.
Such a forum might provide a place where we could continue to benefit
from Pat's involvement, which I have found to be very valuable.
Jonathan
Barry Smith
>I'm going to play Pangloss here, a bit...
>
>In what forum is BFO being critically appraised? As BFO seems to be
>shaping the whole foundry effort, including IAO and OBI, it would seem
>wise to attempt to predict whether it's going to serve the endeavor
>well, what its limits will turn out to be, and whether its reliability
>or applicability might be increased (or threatened) by changes to it.
>This is true even if a sunk cost argument convinces us all not to make
>any changes to it.
>
>Of course I mean an analysis with practical utility, not philosophical
>considerations, as the test (admitting of course that the latter are
>intended to predict the former).
>
>If an appraisal gives BFO high marks, then it should have the benefit
>of bringing skeptics like Pat and me into the fold.
>
>Such a forum might provide a place where we could continue to benefit
>from Pat's involvement, which I have found to be very valuable.
>
>Jonathan
applied in support of scientific (and especially life-science)
research: BFO, DOLCE, SUMO, Upper Level CYC.
The ideal, from Jonathan's point of view, would be to take the same
ontology development problem, have it tackled by four independent
teams of equal competence, using each of these alternative
ontologies, and to measure e.g.:
1. training time taken,
2. quality of resulting ontology,
3. acceptability of resulting ontology to other users,
4. benefits brought by resulting ontology in supporting data
integration, reasoning, etc.
Unfortunately, we do not have the resources to carry out such an
experiment, and we do not have good measures, as yet, for 2.
(conformity e.g. to OWL is by no means a sufficient criterion of
ontology quality) and 4. (it is hard to find a domain which has not
already had ontology resources built, which are now infecting the
data which would be used in the integration; often tilting in a
direction prejudiced in favor of BFO).
However, I would be willing to commit to finding such resources (in
terms of both financing, expertise, and identification of an
independent arbiter) if the proponents of any other top-level
ontology would be willing to match one-on-one in terms of resources
and commitment to a clean experiment.
For the moment, we can note:
i. that matters are not as bad as they seem, given that BFO comes
close to approximating to a sub-ontology of DOLCE, and there is, I
believe, a potential for a comparable partial unification of BFO and SUMO
ii. BFO seems to be of the right size for cross-domain ontology
integration, since it does not overlap with the other sciences and
thus does not overlap with the scientific domain ontologies which
need to be built in integrated fashion
iii. re heading 1., BFO is much smaller than the others, so
(potentially) easier to learn
iv. BFO has demonstrated considerable responsiveness to the needs of
the scientists using it
iv. BFO seems to ahead under heading 3., given that we have some
approximation to a real-world experiment in the sense that people
have been voting with their feet. The list of independent projects
actively using BFO is growing by at least one per week:
http://www.ifomis.org/bfo/users
v. BFO seems to be ahead under a variety of other measures (though
this does not mean that we are satisfied with our performance along
any of these dimensions), including:
A. quality and extent of training resources available,
B. quality and extent of documentation,
C. size and expertise of team working on it, including what
is probably the most sophisticated user community (OBI), which has
already led to major changes in BFO; most of the oddities in current
BFO applications in IAO turn on the fact that IAO is a difficult
domain for which OBI needs a stable representation to deal with a
variety of different kinds of information artifact (I believe that
there is no competing representation of this domain that even comes
close to being able to make the needed distinctions)
D. size and activity of user-group, currently 117 members,
2745 messages
E. responsiveness to critics
F. longevity of engagement on the part of multiple specific users,
G. integration with the RO Relation Ontology, deeply
embedded within the bio-ontology field
(http://genomebiology.com/2005/6/5/R46) with 386 citations on google scholar.
BS
Pat Hayes
On Dec 9, 2009, at 7:53 AM, Jonathan Rees wrote:
> I'm going to play Pangloss here, a bit...
>
> In what forum is BFO being critically appraised?
answer. But there are not many competitors.
> As BFO seems to be
> shaping the whole foundry effort, including IAO and OBI, it would seem
> wise to attempt to predict whether it's going to serve the endeavor
> well, what its limits will turn out to be, and whether its reliability
> or applicability might be increased (or threatened) by changes to it.
> This is true even if a sunk cost argument convinces us all not to make
> any changes to it.
>
> Of course I mean an analysis with practical utility, not philosophical
> considerations, as the test (admitting of course that the latter are
> intended to predict the former).
>
> If an appraisal gives BFO high marks, then it should have the benefit
> of bringing skeptics like Pat and me into the fold.
>
> Such a forum might provide a place where we could continue to benefit
> from Pat's involvement, which I have found to be very valuable.
lurker.
And um... part of my purpose in deciding to withdraw from active
participation was exactly to NOT have this forum get involved in re-
examining the foundations of BFO. If I actively participate, my own
predilictions will, I am quite sure, constantly rub against the BFO
ground assumptions, causing irrelevant frictions in the discussions.
Better to stick with the foundations y'all have chosen, and work
within them, was my point. Please don't understand this as a threat to
re-examine the foundations or else I will go home and take my ball
with me; that would be entirely the wrong message.
By the way, I am pretty sure that the result of, if I you will forgive
the terminology, de-BFOing the final ontology would be largely a
process of simplification at the 'upper' levels of the ontology,
rather than a complete rejection of whatever is produced.
Pat
Barry Smith
Many communities find it useful to hide the top level, and
increasingly sophisticated tools are becoming available for this
purpose, thereby giving rise to the result you here describe. But
those same communities found using BFO useful for building the
ontologies. Thus de-BFOing would be unadvisable, unless an
alternative strategy for building is on hand. Good luck with that.
BS
Bjoern Peters
Alan pointed out to me that my previous email came across as me trying to exclude people from this discussion that don't accept BFO as dogma. That is not the case, and I do want to apologize to all who read it as such.
What I meant to say is that, taken your decision to no longer contribute at this point as a given, it would be great if you could re-join again later to criticize a more complete working draft of IAO. Your contributions have been very valuable, and testing if IAO 'works' from your perspective would be excellent.
Right now, as apparently even Melanie and I can't agree on what a 'concretization' is, we are clearly not yet at a 'working' stage. I don't think we are that far off though, because once we have the 'basic anchoring of IAO in BFO' done and can formulate a set of axioms what an ICE is and how it gets propagated, we should be able to discuss the issues you had raised in the past without referring to high level BFO classes.
Again: apologies if my original email did not convey the above.
Michel Dumontier
Barry Smith
>I would argue that the jargon introduced by the BFO and some related
>ontolgies makes them largely inaccessible to anybody outside of the
>developers. I cannot imagine the challenges in presenting it to the
>layperson. I think people really struggle to understand its
>representation paradigm, and it doesn't help that the theory is
>incomplete. Even through many groups claim to use the BFO (including
>my own), it's unclear how many actually are using it in the way it
>was intended (including my own) - because what little communication
>exists is highly cryptic, or exists in the form of these mailing
>lists or a few slides across several presentations.
organizing classes of an ontology, it is necessary to introduce
strange-sounding distinctions. (Compare the FMA's 'Physical
anatomical entity' and 'Material anatomical entity', the latter a
subtype of the former.) It would not make sense to attempt to
communicate this sort of content to the layperson.
>
>
> i'm not a fan of "well so and so has an ontology of x, so we
> should use it", particularly if it isn't all that clear what it
> means or how it was intended to be used.
> i think part of the reason RO is such a well cited paper is
> because it documents the hard work being done - and is fairly
> clear. more of this has to happen, and certainly the BFO needs to
> be worked out, reviewed, and presented to the people for which it is intended.
>at some point we need to write up what the IAO offers, contrasted
>with other theories - Pat's involvement is critical here, and next
>steps plotted. a scientific approach to ontology design is just as
>important as its philosophical considerations - we should argue for
>innovation, ease of use and utility (not in the sense of, we are
>using it, but in the sense of it has remarkable value and makes
>things that were previously hard, much easier and more coherent).
(with Pat) that it would not make sense to take the opportunity of
clarifying and documenting IAO in a coherent fashion, and redirect it
towards the attempt to go over the many arguments for and against BFO.
BS
>
> > >>
> > >> --
> > >>
> <mailto:informatio...@googlegroups.com>informatio...@googlegroups.com
>
> > >> To change settings, visit
> > >>
> <http://groups.google.com/group/information-ontology>http://groups.google.com/group/information-ontology
>
> > >>
> > >
> > > --
> > >
> <mailto:informatio...@googlegroups.com>informatio...@googlegroups.com
>
> > > To change settings, visit
> > >
> <http://groups.google.com/group/information-ontology>http://groups.google.com/group/information-ontology
> > >
> >
> >------------------------------------------------------------
> >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
> >
> >
> >
> >
> >
> >--
> ><mailto:informatio...@googlegroups.com>information-ontology@
> googlegroups.com
> >To change settings, visit
> ><http://groups.google.com/group/information-ontology>http://groups.
> google.com/group/information-ontology
>--
><mailto:informatio...@googlegroups.com>informatio...@googlegroups.com
>
>To change settings, visit
><http://groups.google.com/group/information-ontology>http://groups.google.com/group/information-ontology
>
>
>
>--
>Michel Dumontier
>Associate Professor of Bioinformatics
>Carleton University
Barry Smith
> > At 08:53 AM 12/9/2009, Jonathan Rees wrote:
> >>If an appraisal gives BFO high marks, then it should have the benefit
> >>of bringing skeptics like Pat and me into the fold.
> >>
> >>Such a forum might provide a place where we could continue to benefit
> >>from Pat's involvement, which I have found to be very valuable.
> >>
> >>Jonathan
> >
> > There are four public domain top-level ontologies which are being
> > applied in support of scientific (and especially life-science)
> > research: BFO, DOLCE, SUMO, Upper Level CYC.
> > The ideal, from Jonathan's point of view, would be to take the same
> > ontology development problem, have it tackled by four independent
> > teams of equal competence, using each of these alternative
> > ontologies, and to measure e.g.:
>
>little evidence that use of a single upper ontology provides substantial
>advantage over the costs that they involve. A base line would need to be
>established.
The control would be the empty ontology, would it not.
Perhaps we have this already in the very impressive mappings work
done by Olivier Bodenreider
<http://mor.nlm.nih.gov:8000/pubs/offi.html>http://mor.nlm.nih.gov:8000/pubs/offi.html
This shows that mappings can be done, even between large
ontology-like resources developed independently, but that they seem
always to be gappy, never completely error-free, very expensive to
build, and even more expensive to maintain.
It is not, I agree, clear that using a single upper ontology would
help. This is why BFO works with the OBO Foundry initiative, and its
orthogonality and cross-product strategy to ensure, as far as
possible, that mappings are not needed.
> > BFO is much smaller than the others, so
> > (potentially) easier to learn
>
>
>smallest complete language, lambda calculus, is extremely hard to learn
>to use.
This may be mistaken. The fact that, in one domain, one small thing
is hard to learn, does not show that in all domains, smaller things
are not easier to learn than larger things.
> > v. BFO seems to be ahead under a variety of other measures (though
> > this does not mean that we are satisfied with our performance along
> > any of these dimensions), including:
> > A. quality and extent of training resources available,
> > B. quality and extent of documentation,
> > C. size and expertise of team working on it, including what
> > is probably the most sophisticated user community (OBI), which has
> > already led to major changes in BFO; most of the oddities in current
> > BFO applications in IAO turn on the fact that IAO is a difficult
> > domain for which OBI needs a stable representation to deal with a
> > variety of different kinds of information artifact (I believe that
> > there is no competing representation of this domain that even comes
> > close to being able to make the needed distinctions)
>
>
>
>the confusion of numeral with number, unit label with unit.
BFO helps to protect from the confusion of numeral with number, and
unit label with unit.
We are working hard to represent the information artifacts in
reality, and to represent the information in reality.
>It's not the only area that BFO does not represent. Energy and force are
>also missing. Processes that change are a problem. It's model of space
>is still problematic.
I agree that there are problems on each of these fronts, but they are
being addressed and we take them seriously.
>Barry is right, BFO has strengths. It could be extended to cope with
>areas for which it is not currently appropriate. But, even given the
>issue above, it's small size is, generally, an advantage.
>
>Stephen Jay Gould used the image of the Burgess shoe horn to describe
>the perils of trying to force experimental data into a classification
>where it does not fit. To my mind, trying to use BFO every would be a
>mistake.
We are attempting to extend it according to need; more will be
revealed when we are sure it works.
BS
>Phil
>
>
>--
>
>You received this message because you are subscribed to the Google
>Groups "BFO Discuss" group.
>To post to this group, send email to bfo-d...@googlegroups.com.
>To unsubscribe from this group, send email to
>bfo-discuss...@googlegroups.com.
>For more options, visit this group at
>http://groups.google.com/group/bfo-discuss?hl=en.
Melanie Courtot
of paper.
My understanding of qualities in BFO (and I may be wrong), is that
they are intrinsic to their bearer. They are "permanent attributes"
that make up the bearer what it is.
Taking the examples in the BFO file: the color of a tomato, the
ambient temperature of air, the circumference of a waist, the shape of
a nose, the mass of a piece of gold, the weight of a chimpanzee. In
all those cases, you can't remove the qualities without destroying the
bearer. Qualities are neither lost or gained, they exist in their
bearer at all time.
In my opinion the fact of having something written on it is not a
quality of a piece of paper. I have several pieces of paper in my
printer that are waiting to be printed on and at the moment don't bear
any information, but they are still pieces of paper. Similarly I would
say that a blank hard drive is still a hard drive.
I think that being an information carrier (to re use the current label
of the class currently in IAO) could be a conferred quality (i.e.
gained as a result of a process, for example writing on a piece of
paper, printing a page, saving information on a hard drive), or
possibly a role played by an independent continuant (a piece of paper
realizes its role information carrier during a reading process for
example, but doesn't realize it when used as cover for the cold air
vent above my desk) - I am unsure about the distinction as this hasn't
really been sorted out yet.
Melanie
On 8-Dec-09, at 4:38 PM, Bjoern Peters wrote:
Barry Smith
At 12:03 PM 12/9/2009, Phillip Lord wrote:
>BFO lacks the ability to represent information. Hence, we end up with
>the confusion of numeral with number, unit label with unit.
ability to represent sugar or water or blood. This is because BFO
represents portions of information (and of sugar, water, blood).
This, in turn, is because BFO rests on the assumption that there is
no information except in the form of portions of information -- which
fall under what we call 'information artifacts', just as there is no
sugar except in the form of portions of sugar.
BFO insists on this because of the many errors people made, and are
still making, when dealing with mass nouns in their ontologies.
BS
Barry Smith
>I am confused about how a pattern of ink can be a quality of a piece
>of paper.
>
>My understanding of qualities in BFO (and I may be wrong), is that
>they are intrinsic to their bearer. They are "permanent attributes"
>that make up the bearer what it is.
>Taking the examples in the BFO file: the color of a tomato, the
>ambient temperature of air, the circumference of a waist, the shape of
>a nose, the mass of a piece of gold, the weight of a chimpanzee. In
>all those cases, you can't remove the qualities without destroying the
>bearer. Qualities are neither lost or gained, they exist in their
>bearer at all time.
qualities which can be gained and lost -- though
all of the above do indeed fall under the heading
of what you call permanent qualities.
Examples of qualities which can be gained and
lost are: headaches, suntans, hunger, ...
There is, though, an issue as to whether the
pattern you mention is a quality of the paper or
a quality of the ink. I think you are right that
it is more correctly described as a quality of
that portion of ink, and in a derivative sense a
quality of the mereological sum of ink and paper.
BS
Pat Hayes
On Dec 9, 2009, at 9:32 AM, Bjoern Peters wrote:
> Pat,
>
> Alan pointed out to me that my previous email came across as me
> trying to exclude people from this discussion that don't accept BFO
> as dogma. That is not the case, and I do want to apologize to all
> who read it as such.
point, which is why I decided to lie low for a while (not completely
vanish, however. :-)
Pat
Bjoern Peters
Barry: It would be nice if BFO could include examples of qualities that are not permanent in the OWL representation. I was aware of the examples you quoted, but having them added to the examples in the OWL representation which Melanie was working off would be very useful. I did not find a BFO tracker. Where could we submit proposals for helpful examples to include in the next version of the BFO OWL representation?
All: I think I have a proposal for the 'ink pattern' quality that should work for all. Note that this is not yet touching on information at all:
'optical surface pattern'=def: a quality inhering in a material entity by virtue of its physical surface makeup that determines reflection and absorption of light.'
This covers other things than ink markings such as pencil writing, laser etchings, collages, etc. Also, every piece of paper has an 'optical surface pattern', including an empty sheet, which has a homogeneous optical surface pattern.
Regarding a hard disk, an 'empty' hard disk still has a magnetization pattern. I would define it as 'a quality of a magnetic material entity in which inheres in its bearer through the orientation of magnetic dipol moments in different parts of the bearer'
For a brain, we just need some kind of analogous property that is not monstrously wrong for neuroscientists. I think we can define 'neural brain pattern' as the pattern of connections and characteristics of the interfaces of neurons in a brain.
I would propose to submit these definitions as 'quality of single physical entity' to PATO.
- Bjoern
> > > Bjoern Peters
> > > Assistant Member
> > > La Jolla Institute for Allergy and Immunology
> > > 9420 Athena Circle
> > > La Jolla, CA 92037, USA
> > > Tel: 858/752-6914
> > > Fax: 858/752-6987
> > > http://www.liai.org/pages/faculty-peters
> >
> >Mélanie Courtot
> >TFL- BCCRC
> >675 West 10th Avenue
> >Vancouver, BC
> >V5Z 1L3, Canada
> >
> >
> >
> >
Robert Hoehndorf
Hi,
>> BFO lacks the ability to represent information. Hence, we end
>> up with the confusion of numeral with number, unit label with
>> unit.
BS> number, and unit label with unit. We are working hard to
BS> represent the information artifacts in reality, and to
BS> represent the information in reality.
As far as I can tell, the main feature of BFO, that helps to make it
applicable in so many domains, is its complete lack of clear criteria
as to what its categories are supposed to represent and how to apply
them within a domain. Consequently, there is nothing that prevents us
from including "number" in BFO:
- as generically dependent continuants that somehow depend on
collections or object aggregates of a certain size; 0 is not
representable that way? No problem, let us make it a defined class
(http://obi-ontology.org/page/Defined_classes) then.
- as specifically dependent continuants (qualities) that are
properties of things of a certain cardinality; similar to its use in
PATO, which has "count" as quality
- as independent continuant; numbers just exist at all times and
everywhere, need no other entities to exist, etc.
- as process boundaries: they exist at every process boundary
- roles: their bearer is the world, and they are in a special "social
set of circumstances" (whatever that means) in which they don't have
to be (whatever that means!) and the world is not changed when a
number ceases to exist (maybe they never cease, so this is trivially
satisfied)
There are probably further viable options, but these may suffice.
However, I don't see how BFO helps to prevent any confusion, not only
the confusion between numeral and number, when the meaning of its
basic categories is hidden behind obfuscated natural language
definition that are open to all kinds of interpretations.
(Example: Disposition:=A realizable entity that essentially (essence?
it is essential for an entity to cause a specific process for it to be
a disposition?) causes (what does "causes" mean? any specific theory
of causation, what are the implications? and where is this relation
mentioned, explicitly, formally described?) a specific process or
transformation in the object (so, a disposition can cause a specific
process (kind of process? an individual process?) or a transformation
in an object (how is this different from a process? if not, why is it
mentioned here?)) in which it inheres, under specific circumstances
(what kind of entity is a circumstance? state of affairs? fact?
situation? a world? And again, what does "specific" mean, does it mean
"individual"?) and in conjunction with the laws of nature (name
one thing that does not happen in "conjunction with" laws of nature!))
>> It's not the only area that BFO does not represent. Energy and
>> force are also missing. Processes that change are a
>> problem. It's model of space is still problematic.
BS> they are being addressed and we take them seriously.
These are not really BFO's problems, I think, as BFO does not seem to
concern itself with energy, force, etc. But the very basic notions in
BFO still remain unclear: what is a dependent continuant in BFO
(explain the "necessary" part of your definition), what is its model
of space, of time, what is a continuant (not a continuant at a time
point, but "the" continuant), how are continuants different from
occurrents (it cannot be their relation to time, because time points
are occurrents; if it is persistence through time, what does this
mean?)?
I don't even mention realizables here, or what "universals" are
supposed to be.
But maybe you do not consider these things (which have been floating
around for years) as problems that are being addressed and taken
seriously...
Rob.
Barry Smith
> >>BFO lacks the ability to represent information. Hence, we end up with
> >>the confusion of numeral with number, unit label with unit.
> >
> > BFO lacks the ability to represent information, just as it lacks the
> > ability to represent sugar or water or blood. This is because BFO
> > represents portions of information (and of sugar, water, blood).
> > This, in turn, is because BFO rests on the assumption that there is
> > no information except in the form of portions of information -- which
> > fall under what we call 'information artifacts', just as there is no
> > sugar except in the form of portions of sugar.
>
>about this portion of zero or that portion of zero doesn't really make
>much sense.
>
>Very simple question. What kind of entity is a number?
BFO is an ontology designed to support data integration in natural
science domains on the basis of the working hypothesis that the best
way to do this is to represent, as coherently as possible, physical
reality. From this perspective, your question is out of scope. (It is
also a very big question, belonging to a very sophisticated
discipline called the philosophy of mathematics; I do not believe
that it will bring benefits to the practical problems facing
bio-ontology to get embroiled in this discipline.)
>My answer, it is an entity in it's own right, with special properties
>which are not the same as the properties of sugar.
But I can agree with this much, at least.
>Properties that we
>understand pretty well from many thousands of years of mathematics.
That's what probably the greatest mind in logic thought when he wrote
his Foundations of Arithmetic. Bertrand Russell, in a 3-line
argument, proved him wrong. I truly don't want to get you embroiled,
but see e.g. here:
http://plato.stanford.edu/entries/frege-logic/
>In short, BFO lacks the ability to represent information because it is
>lacking.
We shall see.
Barry Smith
terms. We will be releasing some better definitions soon; though
these, too, will still surely be imperfect. Natural language is, for
these sorts of purposes, an imperfect vehicle. (But relying on a
formal language has its problems, too.) There is now a great deal of
experience of using BFO, and enough documentation explaining how it
is used, which Robert seems to be ignoring below. His post reminds me
of someone new to physics who says things like:
Well, numbers could be clusters of photons, couldn't they?
BS
Albert Goldfain
> At 07:06 AM 12/10/2009, Phillip Lord wrote:
>
>
>>Barry Smith <phis...@buffalo.edu> writes:
>> > At 12:03 PM 12/9/2009, Phillip Lord wrote:
>> >>BFO lacks the ability to represent information. Hence, we end up with
>> >>the confusion of numeral with number, unit label with unit.
>> >
>> > BFO lacks the ability to represent information, just as it lacks the
>> > ability to represent sugar or water or blood. This is because BFO
>> > represents portions of information (and of sugar, water, blood).
>> > This, in turn, is because BFO rests on the assumption that there is
>> > no information except in the form of portions of information -- which
>> > fall under what we call 'information artifacts', just as there is no
>> > sugar except in the form of portions of sugar.
>>
>>And it is this mistaken assumption that causes the problem. Talking
>>about this portion of zero or that portion of zero doesn't really make
>>much sense.
>>
>>Very simple question. What kind of entity is a number?
>
> BFO is an ontology designed to support data integration in natural
> science domains on the basis of the working hypothesis that the best
> way to do this is to represent, as coherently as possible, physical
> reality. From this perspective, your question is out of scope.
kind of entity is information?". IAO is more useful by ignoring this
question (or better, by letting the user answer it with their favorite
philosophical theory) and by instead focusing on those things in
reality that are the bearers of information or that depend to
information in some way (ICEs, information artifacts, ...) .
Similarly, an ontology of numerical artifacts could be constructed
without ever answering the question "What kind of entity is a number?"
[AG] Subclasses of number change with history: not just the historical
progression of the standard set: naturals, naturals + zero, rationals,
reals, and complex numbers, but also the stranger sorts of numbers
that have come out of mathematics: transfinite cardinal numbers
(Cantor), infinitesimals (Robinson), and quaternions. In fact, the
standard set of numbers are precisely those that find applications in
science (complex numbers, remember, have several applications
describing electricity. What kind of strange universal is supposed to
comprehend all of these?
[AG] Having said all that, we could do a whole lot with BFO to
describe circumstances where numbers and numerals are used (without
confusion!). Supposing I count to three by successively raising my
thumb, raising my index finger, and raising my middle finger. In BFO
terms, my middle finger is bearing the role of the cardinal number
three which is realized during my counting process. There are
problems with this number-as-role account (e.g., my middle finger
could have borne the role of the number one in the counting process if
I had started with it, so it is more of a generic dependence, or
better generic role...but starting with my middle finger would have
been vulgar :-), but the point is that we can in principle come up
with BFO compatible theories for different situations in which numeric
information is conveyed (street address enumeration, credit card
number assignment, ruler measurement, etc.), WITHOUT ever answering
the question "What kind of entity is a number?".
Melanie Courtot
On 9-Dec-09, at 8:35 PM, Bjoern Peters wrote:
> Melanie: I should have been more exact and made clear that I did not
> mean to say that an empty sheet of paper paper had an 'ink pattern'
> quality. Instead, I meant to refer to a quality of what Barry called
> the 'mereological sum of paper and ink'. I think I see now where (at
> least some) of our disconnect came from.
believe this to be independent from this thread so won't expand as
part of this discussion.
Independently of my confusion about "quality", I am happy with saying
the following:
1. having a pattern is a quality of the ink
2. being an information carrier (the class we currently have in IAO)
is a quality of the pattern of ink. It is not a quality of the ink,
but a quality of the pattern formed on the paper when I write
"Vancouver" or the pattern on the "Boston" sign. So that would mean:
((ink has_quality pattern) has_quality information carrier)
Is that correct?
Finally, how will the link between "((ink has_quality pattern)
has_quality information carrier)r" to "(paper on which is written the
ink pattern) has_quality information carrier" be done?
Thanks,
Melanie
Bjoern Peters
1) Every information content entity (ICE) is_concretized_as some (physical) quality. The ICE could be a weight measurement datum. The concretization is a quality such as the optical surface pattern visible on a page of a lab notebook, the magnetization pattern in a hard disk, or the 'neural pattern' in someones brain.
2) Each of these qualities 'inheres in' an independent continuant (the lab notebook page, the hard disk, the brain)
3) There is no 'inheres in' relation between an ICE and an independent continuant other than to the bearers of the concretized ICE qualities
The class 'information bearer' is not currently in IAO, but could be defined with N&S conditions as: material entity and bearer_of some quality and is_concretization_of some ICE. A piece of paper with writing on it would be classified as an information bearer, which answers MCs last question.
I think this is pretty solid.
- Bjoern
Janna Hastings
I see. You want numbers _as forms of information_. That suggests the scope of the IAO. I see IAO already has 'measurement datum', which can be numeric, and have units associated, just as we would need to represent many numeric values in physics or biology. Surely something similar could accommodate calculated numeric values which do not correspond to measurements, something like 'calculated datum' perhaps?
Janna
For example, a calculation process has an output of a number.
Numbers are one of the simplest forms of information, so if we can't do
numbers, then information in general is a problem.
The two options here are straight-forward; either we have numbers
because they are useful or, we don't have them because they introduce
more confusion than their utility. It's a cost-benefit issue.
Phil
Janna Hastings <janna.h...@gmail.com> writes:
> I'm not sure I understand. Why would you like 'number' to be included in
> BFO? Is 'number' a class in an ontology you wish to align with BFO? Or do
> you mean that you wish to include properties such as mass which have numeric
> values?
>
> Thanks, Janna
>
>
Michel Dumontier
Gwen Frishkoff
Dear Bjoern,
I'm trying to catch up on this discussion. I apologize if my question
reflects a basic gap in understanding, and I'd appreciate if you (and
Melanie, Barry and others) can help set me straight.
You have used three examples of ICEs, including a "neural pattern" that
inheres_in (?) someone's brain. I have a few questions about this
example that I will put inline, in response to a portion of your
previous message....
Bjoern Peters wrote:
>>>
>>> All: I think I have a proposal for the 'ink pattern' quality that
>>> should work for all. Note that this is not yet touching on
>>> information at all:
>>>
>>> 'optical surface pattern'=def: a quality inhering in a material
>>> entity by virtue of its physical surface makeup that determines
>>> reflection and absorption of light.'
>>>
>>> This covers other things than ink markings such as pencil
>>> writing, laser etchings, collages, etc. Also, every piece of
>>> paper has an 'optical surface pattern', including an empty sheet,
>>> which has a homogeneous optical surface pattern.
>>>
>>> Regarding a hard disk, an 'empty' hard disk still has a
>>> magnetization pattern. I would define it as 'a quality of a
>>> magnetic
>>
>>> material entity in which inheres in its bearer through the
>>> orientation of magnetic dipol moments in different parts of the
>>> bearer'
>>>
>>> For a brain, we just need some kind of analogous property that is
>>> not monstrously wrong for neuroscientists. I think we can define
>>> 'neural brain pattern' as the pattern of connections and
>>> characteristics of the interfaces of neurons in a brain.
>>>
magnetization_pattern, a neural_pattern is manifestly dynamic
(spatiotemporal). The patterns are defined as changes in levels of
activity over time and space (within and across different regions of the
brain).
Can you clarify how this might be (or is already) handled in BFO and IAO?
How are spatial_patterns_of_(neural)activity represented in BFO/IAO?
How temporal_patterns_of_(neural)activity represented?
In what do these patterns_of_activity inhere? ("brain" is static -- will
that work for temporal_patterns?)
In the abstract (or as data representations), we can think of these
patterns in to ways, either
1) as a sequence of spatial configurations (brain "microstates," where
each microstate represents a network of regions that is "active" with
respect to some baseline); or
2) as a set of time_series, where each time_series represents a temporal
pattern of activity for a single brain microstate.
Are there different BFO representations for 1) and 2)?
I'm looking forward to hearing everyone's thoughts on this.
Thank you,
Gwen
>>> I would propose to submit these definitions as 'quality of single
>>> physical entity' to PATO.
>>>
>>> - Bjoern
>>>
Assistant Professor, Department of Psychology
Associate Member, Neuroscience Institute
140 Decatur Street
Urban Life, Rm 712
Georgia State University
Atlanta, Georgia 30303-3083
Email: gfris...@gsu.edu
Phone: 404-413-6303
Fax: 404-413-6218
"A speaker is like a lousy auto mechanic: every time he fixes something
in the language, he screws up something else." -- Joseph H. Greenberg
Alan Ruttenberg
<philli...@newcastle.ac.uk> wrote:
>
> IAO has, for example, scalar measurement datum which consists of a
> numeral and unit label. So, it covers the printed or visualised
> representation. It does not cover the representation stored on a hard
> drive, for instance.
Measurement unit datum does not use numeral. Michel pointed out in a
previous mail that it uses a datatype property with a float value. Of
course, in OWL floats aren't numbers either ;-)
But we borrow (sort of) numbers from OWL rather than representing them
in IAO explicitly. This is cheating. But useful cheating for the
moment.
You point about alternate representations of numbers it interesting,
though, and deserves some thought. Please consider filing an issue.
http://code.google.com/p/information-artifact-ontology/issues/list
> Likewise, it has "conclusion textual entity"; conclusion is apparently,
> like number, hard, but text about it is okay.
We're more able to identify the textual entities and be clear about
identity criteria for them. Still waiting for a good proposal on how
to handle conclusions in general. I haven't been impressed by any so
far.
What's your strategy for dealing with hard things? My worry is that a
common way of dealing with hard things is to add a term to the
ontology with a label and no definition, and hope for the best. I
haven't found that to work very well.
> IAO demonstrates the problem with BFO. Barry argues that BFO clearly
> differentiates between numbers and numerals; I say, that by considering
> numbers out-of-scope, we are left with representing numerals instead, so
> BFO is actually forcing us to conflate the two.
Do you have an alternative that we should look at for inspiration?
> I think that IAO is less clear than it should be as a result.
Well, it's hard to argue that it couldn't use more clarity. But
frankly, I think IAO even as it is has made progress over what already
exists, and I credit BFO for bringing a perspective that I've found
helpful.
You said: "Numbers are one of the simplest forms of information".
Perhaps this makes sense to you, but I don't happen to find it very
illuminating. What definition of information supports the idea that
number is a subclass of it?
Elsewhere you say: "Ontologies should represent our shared confusion
as well as our shared understanding." How is representing our shared
confusion going to make IAO more clear?
-Alan
Barry Smith
>On Dec 17, 2009, at 9:13 PM, Barry Smith wrote:
>
>>At 10:02 PM 12/13/2009, you wrote:
>>
>>>On Dec 12, 2009, at 10:47 AM, Barry Smith wrote:
>>>
>>>>At 01:59 AM 12/11/2009, you wrote:
>>>>
>>>>>On Dec 10, 2009, at 8:15 AM, Barry Smith wrote:
>>>>>
>>>>>>As I say, we are doing our best to improve the definitions of BFO
>>>>>>terms. We will be releasing some better definitions soon; though
>>>>>>these, too, will still surely be imperfect. Natural language is,
>>>>>>for
>>>>>>these sorts of purposes, an imperfect vehicle. (But relying on a
>>>>>>formal language has its problems, too.) There is now a great
>>>>>>deal of
>>>>>>experience of using BFO, and enough documentation explaining how
>>>>>>it
>>>>>>is used
>>>>>
>>>>>Can you briefly point to some of this? That would be genuinely
>>>>>helpful
>>>>>for me.
>>>>
>>>>You can start digging around here:
>>>>http://www.ifomis.uni-saarland.de/bfo/
>>>>The 'manual' was put together in haste be a student of mine some
>>>>time ago, so treat it gentlily. We are preparing a thoroughly
>>>>revised version for MIT Press, and I may be able to send you the MS
>>>>in the near future. There is more , but this is probably enough to
>>>>get yous started.
>>>
>>>Thanks, but I was mostly wondering about papers reporting the
>>>*experience of using* you refer to above.
>>
>>There are several links under the users tab at the
>>
>>http://www.ifomis.uni-saarland.de/bfo/
>>
>>site
>
>OK, thanks.
>
>>
>>>Do you actually believe all this stuff about universals comprising
>>>reality, and ontologies being solely concerned with the description
>>>of
>>>universals? (Your student needs to read more about grue and bleen, by
>>>the way :-)
>>
>>I actually do believe it, though it is clear that we need facility
>>for ontology terms representing classes which are not extensions of
>>universals, in addition to the first-class terms representing
>>universals. The latter is, I'm afraid, what science is all about,
>>and that your talk of 'properties' as capturing what is general in
>>reality can properly be made sense of only by appeal to universals.
>
>It seems increasingly clear that you and I live in entirely different
>worlds. I really, really have absolutely no idea what you are talking
>about. And I do know quite a lot about (some) sciences :-)
>
>What I find so strange about much of these emails is that people quite
>frequently express puzzlement about whether something is a continuant,
>or is a universal, and you infallibly tell them the answer, but you
>never tell them what criteria you are using to make these decisions.
>Your recent reply to P.Def, for example, says that Fur is 'close to'
>being a universal, but that 'furry animal' is not. But Barry, *how* do
>you make such judgements? What thoughts or criteria are behind them?
>To me, they sound like completely arbitrary pronouncements. I have no
>idea how you are making them, even intuitively. If I were shown some
>other examples, the only way I could possibly answer such questions
>would be to ask you.
I have been trying to teach you, but I seem not to be doing a very good job.
First, being a universal is like being true; there is no easy marker,
so that we can point, e.g. to pieces of language and say: these
represent universals, those do not. There are some linguistic clues
(universals are likely not referred to by means of disjunctive
noun-phrases, for example), but the only effective way to discover
what the universals are in reality is to do science, to see what
terms are used indispensably in the formulation of scientific laws.
Nowadays, some of this science is done with the help of ontology
building. All of it is, like all science, tentative and provisional;
but also (I hope we are all sensible enough to believe) getting
better with time.
Biologists engaged in collecting and reasoning with phenotype
descriptions use terms like 'furry' to describe phenotypes. This
leads me to hypothesize that 'furriness' designates a universal. (We
have lots of instances; the term is used, at least, in scientific
descriptions.)
Does 'furry animal' represent a universal? I believe not. But
fortunately we do not need to worry about this, since we have the two
universals, furriness and animal, and we can build ontologies by
using cross-products without having to commit ourselves to the
existence of the furry animal universal in their asserted hierarchies.
>>(Properties alone do not capture what is general, since we have
>>properties like 'is identical to Patrick Hayes')
>
>Generality is expressed using quantifiers.
not all by themselves, it isn't
>But there is no distinction
>in kind between a property which is true for large swathes of reality
>and one which it true only very locally. They are both properties, but
>differ in extent.
See my remarks on what you call 'kinds' in earlier email.
BS
Pat Hayes
The reason is probably because I have long since decided that the
distinction between universal and not universal, or if you prefer, the
category of 'universal', is ultimately meaningless. I say,
ultimately, because of course it - the distinction - may be reasonably
clear in a given context of use. But I do not believe that it is a
sensible distinction to insist upon in any ontology that is likely to
be extended or to interact with other ontologies (as the status of
'universal' is likely to change with context), and also that it has no
ultimate meaning (because what counts as 'universal' is going to
change as science changes, or even within specialized subdisciplines)
and also that it has no real utility (since to count something as not
universal says nothing about it other than that it can admit
exceptions, which is true of virtually all properties that have ever
been conceptualized by any thinker.) It seems to me that you in fact
agree with this, in essence. Your own judgements about universality
seem to often be ad-hoc, guarded or provisional.
> First, being a universal is like being true; there is no easy marker,
> so that we can point, e.g. to pieces of language and say: these
> represent universals, those do not. There are some linguistic clues
> (universals are likely not referred to by means of disjunctive
> noun-phrases, for example), but the only effective way to discover
> what the universals are in reality is to do science, to see what
> terms are used indispensably in the formulation of scientific laws.
Quite. But surely the lesson of science is that the only real
universals are to be found in particle physics, and even those are
under constant review and hostage to theoretical advances which are
universally acknowledged to be incomplete.
> Nowadays, some of this science is done with the help of ontology
> building. All of it is, like all science, tentative and provisional;
> but also (I hope we are all sensible enough to believe) getting
> better with time.
>
> Biologists engaged in collecting and reasoning with phenotype
> descriptions use terms like 'furry' to describe phenotypes. This
> leads me to hypothesize that 'furriness' designates a universal. (We
> have lots of instances; the term is used, at least, in scientific
> descriptions.)
>
> Does 'furry animal' represent a universal? I believe not. But
> fortunately we do not need to worry about this, since we have the two
> universals, furriness and animal, and we can build ontologies by
> using cross-products without having to commit ourselves to the
> existence of the furry animal universal in their asserted hierarchies.
We do not have to commit to it AT ALL. We can simply write ontologies
in logic, completely ignoring this question of what counts as a
universal. As I write axioms I never seem to have to ask myself, but
is this a universal?? The question simply does not arise. So let us
ignore it. We have enough problems already, without inventing
unanswerable questions for ourselves to answer.
Pat
>
>>> (Properties alone do not capture what is general, since we have
>>> properties like 'is identical to Patrick Hayes')
>>
>> Generality is expressed using quantifiers.
>
> not all by themselves, it isn't
>
>> But there is no distinction
>> in kind between a property which is true for large swathes of reality
>> and one which it true only very locally. They are both properties,
>> but
>> differ in extent.
>
> See my remarks on what you call 'kinds' in earlier email.
>
> BS
>
> --
> informatio...@googlegroups.com
> To change settings, visit
> http://groups.google.com/group/information-ontology
>
------------------------------------------------------------
Barry Smith
This is why we should never buy a laptop, since the laptops they will
be selling next month will be better and cheaper.
>
>and also that it has no real utility (since to count something as not
>universal says nothing about it other than that it can admit
>exceptions, which is true of virtually all properties that have ever
>been conceptualized by any thinker.)
I believe that we are beginning to understand how to deal with this,
in biology at least, through the ideas on canonicity introduced by Rosse
http://www.ncbi.nlm.nih.gov/pubmed/14759820
and drawing on the idea, already present in Aristotle, that
universals (natural kinds) always have borderline instances.
But note that the crucial distinction is not that between universals
and ad hoc classes, but that between universals and instances.
>It seems to me that you in fact
>agree with this, in essence. Your own judgements about universality
>seem to often be ad-hoc, guarded or provisional.
>
> > First, being a universal is like being true; there is no easy marker,
> > so that we can point, e.g. to pieces of language and say: these
> > represent universals, those do not. There are some linguistic clues
> > (universals are likely not referred to by means of disjunctive
> > noun-phrases, for example), but the only effective way to discover
> > what the universals are in reality is to do science, to see what
> > terms are used indispensably in the formulation of scientific laws.
>
>Quite. But surely the lesson of science is that the only real
>universals are to be found in particle physics,
sounds like the no true Scotsman fallacy:
Teacher: All Scotsmen enjoy haggis.
Student: My uncle is a Scotsman, and he doesn't like haggis!
Teacher: Well, all true Scotsmen like haggis.
> and even those are
>under constant review and hostage to theoretical advances which are
>universally acknowledged to be incomplete.
Let me repeat, for the 9th and last time, that I am aware that
science is subject to correction, and that those ontologies which are
parts of science will therefore also be subject to correction.
> > Nowadays, some of this science is done with the help of ontology
> > building. All of it is, like all science, tentative and provisional;
> > but also (I hope we are all sensible enough to believe) getting
> > better with time.
> >
> > Biologists engaged in collecting and reasoning with phenotype
> > descriptions use terms like 'furry' to describe phenotypes. This
> > leads me to hypothesize that 'furriness' designates a universal. (We
> > have lots of instances; the term is used, at least, in scientific
> > descriptions.)
> >
> > Does 'furry animal' represent a universal? I believe not. But
> > fortunately we do not need to worry about this, since we have the two
> > universals, furriness and animal, and we can build ontologies by
> > using cross-products without having to commit ourselves to the
> > existence of the furry animal universal in their asserted hierarchies.
>
>We do not have to commit to it AT ALL. We can simply write ontologies
>in logic, completely ignoring this question of what counts as a
>universal.
When people attempt to build ontologies following this rule, they
often make errors some of which could be avoided if they are taught
to avoid confusing universals and instances.
http://www.bioontology.org/top-10-errors
is just a sampling from lists of thousands of such errors which I can
supply you with if you insist.
> As I write axioms I never seem to have to ask myself, but
>is this a universal?? The question simply does not arise. So let us
>ignore it. We have enough problems already, without inventing
>unanswerable questions for ourselves to answer.
You really should get out more, and discover how hard it is for those
who were not trained in philosophy to avoid use-mention confusions,
is-a overloading, confusing knowledge with the thing known, confusing
information with the object that it is information about, and all the
other things which generate crappy ontologies, some of them happily
imposed e.g. by the Federal Government on US hospitals.
http://ontology.buffalo.edu/hl7/doublestandards.pdf
BS
Pat Hayes
But laptops, unlike pointless distinctions, are useful.
>> and also that it has no real utility (since to count something as not
>> universal says nothing about it other than that it can admit
>> exceptions, which is true of virtually all properties that have ever
>> been conceptualized by any thinker.)
>
> I believe that we are beginning to understand how to deal with this,
> in biology at least, through the ideas on canonicity introduced by
> Rosse
> http://www.ncbi.nlm.nih.gov/pubmed/14759820
> and drawing on the idea, already present in Aristotle, that
> universals (natural kinds)
Whoa. A universal is a natural kind? That seems like a new/different
idea. I thought universals were supposed to be *universal*. Natural
kinds are a much messier notion (and probably more useful.) Whether
something is a natural kind isn't really of central importance for
reasoning, for example, but it can avoid pointlessly looking for a
definition.
> always have borderline instances.
> But note that the crucial distinction is not that between universals
> and ad hoc classes, but that between universals and instances.
No, thats is the set/member distinction, which applies both to
properties that are universals and those that are not.
>> It seems to me that you in fact
>> agree with this, in essence. Your own judgements about universality
>> seem to often be ad-hoc, guarded or provisional.
>>
>>> First, being a universal is like being true; there is no easy
>>> marker,
>>> so that we can point, e.g. to pieces of language and say: these
>>> represent universals, those do not. There are some linguistic clues
>>> (universals are likely not referred to by means of disjunctive
>>> noun-phrases, for example), but the only effective way to discover
>>> what the universals are in reality is to do science, to see what
>>> terms are used indispensably in the formulation of scientific laws.
>>
>> Quite. But surely the lesson of science is that the only real
>> universals are to be found in particle physics,
>
> sounds like the no true Scotsman fallacy:
>
> Teacher: All Scotsmen enjoy haggis.
> Student: My uncle is a Scotsman, and he doesn't like haggis!
> Teacher: Well, all true Scotsmen like haggis.
>
>> and even those are
>> under constant review and hostage to theoretical advances which are
>> universally acknowledged to be incomplete.
>
> Let me repeat, for the 9th and last time, that I am aware that
> science is subject to correction, and that those ontologies which are
> parts of science will therefore also be subject to correction.
I know, but you also insist upon ignoring the engineering consequences
of this for ontology construction. If you know that something is going
to change, or is going to be different in the next ontology over (the
one built for the other subdiscipline), and if its not particularly
useful or important in any case, why insist on giving it such
importance and a name like "universal" ?
>
>>> Nowadays, some of this science is done with the help of ontology
>>> building. All of it is, like all science, tentative and provisional;
>>> but also (I hope we are all sensible enough to believe) getting
>>> better with time.
>>>
>>> Biologists engaged in collecting and reasoning with phenotype
>>> descriptions use terms like 'furry' to describe phenotypes. This
>>> leads me to hypothesize that 'furriness' designates a universal. (We
>>> have lots of instances; the term is used, at least, in scientific
>>> descriptions.)
>>>
>>> Does 'furry animal' represent a universal? I believe not. But
>>> fortunately we do not need to worry about this, since we have the
>>> two
>>> universals, furriness and animal, and we can build ontologies by
>>> using cross-products without having to commit ourselves to the
>>> existence of the furry animal universal in their asserted
>>> hierarchies.
>>
>> We do not have to commit to it AT ALL. We can simply write ontologies
>> in logic, completely ignoring this question of what counts as a
>> universal.
>
> When people attempt to build ontologies following this rule, they
> often make errors some of which could be avoided if they are taught
> to avoid confusing universals and instances.
I grant you that people make errors. I am far less convinced that
teaching them to avoid confusing warbles with blongles is going to
help them avoid errors.
> http://www.bioontology.org/top-10-errors
> is just a sampling from lists of thousands of such errors which I can
> supply you with if you insist.
>
>> As I write axioms I never seem to have to ask myself, but
>> is this a universal?? The question simply does not arise. So let us
>> ignore it. We have enough problems already, without inventing
>> unanswerable questions for ourselves to answer.
>
> You really should get out more, and discover how hard it is for those
> who were not trained in philosophy to avoid use-mention confusions,
> is-a overloading, confusing knowledge with the thing known, confusing
> information with the object that it is information about, and all the
> other things which generate crappy ontologies, some of them happily
> imposed e.g. by the Federal Government on US hospitals.
I do get out quite a lot, and the moral I draw from all this is that
philosophers should make less noise and be more humble. Have you
noticed that in spite of all these deplorable philosophical errors,
the world actually manages to carry on quite well in getting its
business done? That the ubiquity of use/mention confusions in language
since time immemorial (I noticed a humdinger recently while listening
to the Messiah: "and his name shall be called, wonderful counselor,
the everlasting father, the prince of peace.") does not, in fact, seem
to greatly matter. People do not confuse the word for the world, in
spite of the way that they talk. Nor do they confuse objects with
thoughts, etc.. How is it that apparently foolish errors like these
don't cause the fabric of societal communication to rend apart? There
is a moral here for our own professional work, and its not that a few
enlightened souls should take on the task of teaching the rest of the
planet how to Think Proper.
But apart from this larger issue, and even if we were to accept that
getting ontologies damn-silly-error-free is a good goal to attempt,
its not at all clear how adopting this particular meaningless
distinction (between universal and mere property) is going to help
avoid making use/mention confusions. They seem on the face of it to
have nothing to do with one another.
Pat
Barry Smith
I invite you to compare the current version of the Gene Ontology,
above all its structure and definitions, with the way it was before I
became involved (when it contained nice definitions such as hemolysis
=def. the causes of hemolysis). The Gene Ontology itself is
demonstrably useful. The Gene Ontology Consortium demonstrably finds
what I do useful.
> >> and also that it has no real utility (since to count something as not
> >> universal says nothing about it other than that it can admit
> >> exceptions, which is true of virtually all properties that have ever
> >> been conceptualized by any thinker.)
> >
> > I believe that we are beginning to understand how to deal with this,
> > in biology at least, through the ideas on canonicity introduced by
> > Rosse
> > http://www.ncbi.nlm.nih.gov/pubmed/14759820
> > and drawing on the idea, already present in Aristotle, that
> > universals (natural kinds)
>
>Whoa. A universal is a natural kind?
as I explained earlier in this interchange, e.g. on Dec. 18. And
again, one reason for using 'universal' rather than 'type', 'kind',
'sort', etc., is that, because it is so ugly, it has not become
associated with a detritus of alien meanings.
>That seems like a new/different
>idea. I thought universals were supposed to be *universal*. Natural
>kinds are a much messier notion (and probably more useful.) Whether
>something is a natural kind isn't really of central importance for
>reasoning, for example, but it can avoid pointlessly looking for a
>definition.
>
> > always have borderline instances.
> > But note that the crucial distinction is not that between universals
> > and ad hoc classes, but that between universals and instances.
>
>No, thats is the set/member distinction, which applies both to
>properties that are universals and those that are not.
You still, I'm afraid, don't get it.
In the 'natural kind' idiom you seem to prefer, there is something of
a logically different sort being said when we assert
[1] Fido is an instance of the natural kind dog
and
[2] Fido is a member of the set {Fido, 3, the planet Jupiter}
And again (since we have been round this circle at least once
already), if we say
[3] Fido is a member of the set of dogs
then what we are saying is really
[4] Fido is a member of the set defined by the fact that all its
members are instances of the natural kind dog
and so we cannot understand [4] unless we already understand [1].
Not at all. I take these consequences very seriously, as my ontology
engineering colleagues will attest.
>If you know that something is going
>to change, or is going to be different in the next ontology over (the
>one built for the other subdiscipline), and if its not particularly
>useful or important in any case, why insist on giving it such
>importance and a name like "universal" ?
The exquisite pleasures gained by irritating you by using this name
are, as they say, priceless.
It did and it does. The full strategy is a work in progress, and we
are learning all the time about how to do it, but it is having
demonstrable positive effects already -- and no one has been able to
work out a competing strategy which has equivalent positive effects.
> > http://www.bioontology.org/top-10-errors
> > is just a sampling from lists of thousands of such errors which I can
> > supply you with if you insist.
> >
> >> As I write axioms I never seem to have to ask myself, but
> >> is this a universal?? The question simply does not arise. So let us
> >> ignore it. We have enough problems already, without inventing
> >> unanswerable questions for ourselves to answer.
> >
> > You really should get out more, and discover how hard it is for those
> > who were not trained in philosophy to avoid use-mention confusions,
> > is-a overloading, confusing knowledge with the thing known, confusing
> > information with the object that it is information about, and all the
> > other things which generate crappy ontologies, some of them happily
> > imposed e.g. by the Federal Government on US hospitals.
>
>I do get out quite a lot, and the moral I draw from all this is that
>philosophers should make less noise and be more humble.
I agree.
>Have you
>noticed that in spite of all these deplorable philosophical errors,
>the world actually manages to carry on quite well in getting its
>business done? That the ubiquity of use/mention confusions in language
>since time immemorial (I noticed a humdinger recently while listening
>to the Messiah: "and his name shall be called, wonderful counselor,
>the everlasting father, the prince of peace.") does not, in fact, seem
>to greatly matter.
http://www.syleum.com/2009/03/17/healthcare-data-model/
> People do not confuse the word for the world, in
>spite of the way that they talk.
Computers, unfortunately, do. The members of many healthcare
standards bodies (etc.) do. The authors of W3C standards documents do.
> Nor do they confuse objects with
>thoughts, etc.. How is it that apparently foolish errors like these
>don't cause the fabric of societal communication to rend apart?
An example (it would be funnier if the HL7 organization (represented
here by Dan) were not involved in keeping us alive when we get to hospital):
http://hl7-watch.blogspot.com/2006/02/is-there-difference-between-person-and.html
I can give you the names, off-line, of leading healthcare IT
professionals who have been tearing their hair out for years because
of the incoherence of the HL7 standard along precisely the dimensions
you, whistling Haendel, see no problems with.
>There
>is a moral here for our own professional work, and its not that a few
>enlightened souls should take on the task of teaching the rest of the
>planet how to Think Proper.
Again, there is a significant well-authenticated demand for people
who can take on this task.
>But apart from this larger issue, and even if we were to accept that
>getting ontologies damn-silly-error-free is a good goal to attempt,
>its not at all clear how adopting this particular meaningless
>distinction (between universal and mere property) is going to help
>avoid making use/mention confusions. They seem on the face of it to
>have nothing to do with one another.
The confusion of property with predicate is one example of the
use-mention confusion.
I have argued earlier in this interchange that, to get clear about
the relations between predicates and properties we need to become
clear about the fact that not all predicates correspond to
properties. The reason is analogous to the reason why [1] and [2]
above are logically distinct sorts of assertions. You get one clear,
there is hope that you can get the other clear.
BS
Pat Hayes
On Dec 28, 2009, at 11:29 AM, Barry Smith wrote:
....
>>>
>>> This is why we should never buy a laptop, since the laptops they
>>> will
>>> be selling next month will be better and cheaper.
>>
>> But laptops, unlike pointless distinctions, are useful.
>
> I invite you to compare the current version of the Gene Ontology,
> above all its structure and definitions, with the way it was before I
> became involved (when it contained nice definitions such as hemolysis
> =def. the causes of hemolysis). The Gene Ontology itself is
> demonstrably useful. The Gene Ontology Consortium demonstrably finds
> what I do useful.
Im sure you do many useful things, Barry. But I was referring to the
notion of Universal, not to you.
Well, Im afraid I still disagree. These are not different in their
*logic*. They differ, perhaps, in that the sets in question differ.
But even this is very hard to make into a *logical* distinction. In
both cases, the logic (in the exact sense of the set of correct
entailments from the assertion) sanctions exactly the same kinds of
entailments. The logic does not care about what 'kind' of set - or, if
you prefer, what kind of predication - it is, only that it is a
predication.
> And again (since we have been round this circle at least once
> already), if we say
>
> [3] Fido is a member of the set of dogs
>
> then what we are saying is really
>
> [4] Fido is a member of the set defined by the fact that all its
> members are instances of the natural kind dog
True. But one can say this about any set, membership in which can be
otherwise characterized linguistically. X is a member of the set {x:
P(x) } just when P(X).
>
> and so we cannot understand [4] unless we already understand [1].
Not at all. They are just two ways of saying the same thing. Neither
of them has epistemic priority over the other. True, the 'set' way of
talking may be less natural to someone without a mathematical
background; but then I have never argued that we should be talking of
sets explicitly. I prefer the simple FO style of predication, using
CLIF notation: (Dog Fido), (Silly Fido) , where someone has, for some
crazy reason, asserted that
(forall (x)(iff (Silly x)(or (= x Fido)(= x 3)(= x Jupiter)) ))
Silly is, of course, silly; but once it has been introduced into the
lexicon, (Silly Fido) is perfectly meaningful and has the same
relation to Silly (and to Fido) that (Dog Fido) has to Dog (and Fido).
Pat
PS I will not pursue the rest of this exchange, as it is getting way
too ad hominem to be useful. You may have the last word.
Barry Smith
flogged this to death.
BS