Qualities of processes

[Chris Mungall]

Can someone remind me why qualities of processes (e.g. rates) are disallowed? Is there documentation e.g. on a wiki somewhere?

Let's just say that BFO were to allow qualities to inhere in processes. There would be many pragmatic advantages. Modeling would be easier and more in line with scientists' intuitions.

What bad things would happen? Would interoperability in some way be hindered? Am I missing something?

[Barry Smith]

The argument, all along, is that when we make predications concerning processes we are not referring to some extra entity, as we are when we assert, for example, that a continuant has a suntan or has a disease; rather we are asserting that the process is of a certain type, for instance it is of the type beating process, or of the type beating process with rate 60bpm, and so forth.

We have finally (I hope) found a way to document this idea coherently, and I hope that it will be in good enough shape to present at the meeting on November 10-11.

BS

--
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.

[Robert Hoehndorf]

>>>>> "BS" == Barry Smith <phis...@buffalo.edu> writes:

BS> The argument, all along, is that when we make predications
BS> concerning processes we are not referring to some extra entity,
BS> as we are when we assert, for example, that a continuant has a
BS> suntan or has a disease; rather we are asserting that the
BS> process is of a certain type, for instance it is of the type
BS> beating process, or of the type beating process with rate 60bpm,
BS> and so forth.

The same choice can be made for objects: the apple's red does not
require an additional entity of the type "red" that inheres in the
apple, but rather the apple could just be of the type "red" (or "red
thing"), or it could just be of the type "red, round thing with 10cm
diameter", etc.

You have not answered the question about why BFO makes that particular
choice, rather reiterated the consequences of the choice. And the
consequences you describe are what makes the use of BFO problematic,
since they are both counterintuitive and incompatible with the existing
database/ontology designs.

R

[Barry Smith]

On Tue, Oct 25, 2011 at 8:43 AM, Robert Hoehndorf <leec...@leechuck.de> wrote:

>>>>> "BS" == Barry Smith <phis...@buffalo.edu> writes:

   BS> The argument, all along, is that when we make predications
   BS> concerning processes we are not referring to some extra entity,
   BS> as we are when we assert, for example, that a continuant has a
   BS> suntan or has a disease; rather we are asserting that the
   BS> process is of a certain type, for instance it is of the type
   BS> beating process, or of the type beating process with rate 60bpm,
   BS> and so forth.

The same choice can be made for objects: the apple's red does not
require an additional entity of the type "red" that inheres in the
apple, but rather the apple could just be of the type "red" (or "red
thing"), or it could just be of the type "red, round thing with 10cm
diameter", etc.

Thing owned by the emporer could be of the type "thing owned by the emporer" I guess.

 

You have not answered the question about why BFO makes that particular
choice, rather reiterated the consequences of the choice. And the
consequences you describe are what makes the use of BFO problematic,
since they are both counterintuitive and incompatible with the existing
database/ontology designs.

Please be patient. We hope to have an account which will meet the practical need soon. 

BS

[Ludger Jansen]

Robert, maybe an analogy can help:

A pizza with a salty taste is a pizza with a lot of salt in it. However, a pizza with an intensive taste is not a pizza with a lot of intensivity in it!
Likewise, an apple needs the property of redness as "an additional entity" to be red -- for the apple itself could also be green, thus as such it does not account for its being red.

But if you have a process, things are different. My going from A to B is a process. What makes this a fast process? Do we have to add some fastness to the process? No, it's simply that I have to go quicker. It's not an additional property belonging to a process, but ME changing in a different way.

What makes this problematic and counterintuitive? It's not, I think, the ontological analysis, but the restrictions it implies for getting expressivity on the simple. Natural language allows us to make just anything subject of a predication. But this is a grammatical issue, which probably nobody disputes. It does not mean that just anything is bearer of a property (which is an ontological issue).

Best
Ludger

--
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.

-- 

PD Dr. Ludger Jansen
Institut für Philosophie
Universität Rostock
18051 Rostock

[Adam M. Goldstein]

On Oct 25, 2011, at 8:43, Robert Hoehndorf <leec...@leechuck.de> wrote:

>>>>>> "BS" == Barry Smith <phis...@buffalo.edu> writes:
>
> BS> The argument, all along, is that when we make predications
> BS> concerning processes we are not referring to some extra entity,
> BS> as we are when we assert, for example, that a continuant has a
> BS> suntan or has a disease; rather we are asserting that the
> BS> process is of a certain type, for instance it is of the type
> BS> beating process, or of the type beating process with rate 60bpm,
> BS> and so forth.
>
> The same choice can be made for objects: the apple's red does not
> require an additional entity of the type "red" that inheres in the
> apple, but rather the apple could just be of the type "red" (or "red
> thing"), or it could just be of the type "red, round thing with 10cm
> diameter", etc.
>

Interesting discussion.

Sometimes some people [1] will say that it's the determinates, not the determinables, that are real. So, it's that red is real (determinate, ie, what color the thing is) while color is not, because it's not specific enough, ie, a thing can have a given color, which requires further determination.

I suppose on this view, red is a kind of color, and red cubes are kinds of cubes. There are 60 bpm beating processes, 61 bpm beating processes, etc.

The universals are real, eg red, 60, but in the world as it is, there are only instances of them that appear together with apples etc and beating processes etc.

All the things of which there are 60 are presumably not interesting, as a group, but the beating processes in their various determinations are (or at least are tied to other universals like probability of cardiac arrest within n units of time, a connection we'd really care about).

How do the official BFO rules compare with this determinate/determinables view? I don't have that view or advocate it, but I think it's a good place to start.

> You have not answered the question about why BFO makes that particular
> choice, rather reiterated the consequences of the choice.

I agree---in the initial response by Barry, an earlier message in this thread, the answer just seems to be, wait until we present a paper on this in November.

Maybe abstracting the view or posting a draft of the paper would help. In fact I am sure it would!

Adam

[Melissa Haendel]

Hi all,

I thought a little scientists perspective might be helpful....

I need to be able to represent that the rate of a process in one organism is decreased relative to the rate of the process in another organism. 

eg. brain development is occurring at a reduced rate in a shh -/- zebrafish mutant, relative to a wildtype sibling of the AB strain.

I also need to be able to represent that a process has a change in onset relative to another process. Some examples are:

chondrogenesis has a late onset relative to normal (relative to the development of the organism as a whole)

chondrogenesis has a late onset relative to osteogenesis (one processs is delayed in its start relative to another)

I need to be able to describe a process occurring asynchronously.

for example, asynchronous pigmentation.

Scientists need to be able to record these aspects of processes as part of phenotype annotation. We don't care what we call them, we just need to be able compute on them. For example, find me all organisms that have asynchrony in all neural crest development processes. 

I hope that some of these examples (and the many others that I am sure you have considered) are helpful in making BFO work for this purpose. Scientists need a straightforward mechanism to represent process phenotypes, and ontologists need a way to compute on them. We need to ensure that whatever is done in BFO to this end isn't so complex so as to deter the scientists from recording these very important aspects of phenotype or we will not get the data that we need to do our informatics work.

Best,

Melissa

Dr. Melissa Haendel

eagle-i Networking Research Resources

OHSU Library http://www.ohsu.edu/library/

Department of Medical Informatics and Epidemiology
Oregon Health & Science University
hae...@ohsu.edu
skype: melissa.haendel

503-407-5970

[Pierre Grenon]

On Mon, Oct 24, 2011 at 9:49 PM, Chris Mungall <cjmu...@lbl.gov> wrote:
> Can someone remind me why qualities of processes (e.g. rates) are disallowed? Is there documentation e.g. on a wiki somewhere?

doctrinally, they don't exist. the justification is philosophical
which is the accepted justification mechanism according to BFO's
methodology. Ain't that, perhaps swiftly, dealt with in some paper?

practically, in a knowledge representation based on BFO, they are
abstractions. Terms referring to them could be introduced in a
knowledge representation modulo the understanding they are not
referring to anything in reality. I don't see what you would gain with
classes of this sort and it seems more straightforward to have a
property that maps processes to something similar to a conceptual
space (which BFO accept as a knowledge representation extension but
with which BFO is not concerned in ontological terms).

> Let's just say that BFO were to allow qualities to inhere in processes.

BFO would then need to account for the theory of such 'inherence'.
There is something misleading in the assumption. While inherence
already has a precise technical meaning in BFO, the assumption seems
to be that inherence is the relation that would obtain between process
attributes/qualities and processes. This assumption seems to be
relying on the superficial linguistic features, predication in
English, in particular. These are not necessarily to be brushed aside
but they are, methodologically speaking, without compelling
consequences for the ontology.

It's similar to another common sensical move suggested long ago, i.e.
have an overarching category of regions in BFO, because, well, the
classes of spatial regions and of spatiotemporal regions are both
named after some type of regions, ain't they? In the present
methodlogical context, this is not a compelling argument, nor more
than to say that there is an overarching category of ducks, ducks and
rubber ducks.

> There would be many pragmatic advantages.

I think the advantages you are expecting are tied to having a
knowledge representation with an expressive vocabulary. This is
arguably not the same thing as having an ontology.

From a knowledge representation standpoint, you would gain the
adeqaute expressivity without a category of process qualities but with
the formal language for mapping processes to a conceptual space of
sort. Arguably, that would also allow for a more insightful definition
of some class of alleged property of process.

> Modeling would be easier and more in line with scientists' intuitions.

Scientists have no intuitions that processes have properties in an
ontological sense.

Scientists have a need to describe phenomena in certain ways and
perhaps to use a language allowing to attribute certain properties (in
a logical and mathematical sense) to certain variables. This is
predication in the broadest.

The idea of ontologies is to find out what grounds predication, it is
not to introduce syntactical elements to reproduce predication, which
would make ontological engineering hardly more than a cumbersome
translation exercise. The distinction makes sense when abiding by
BFO's methodlogical principles and motivations.

Nevertheless, you can develop the knowledge representation in a useful
way (e.g. for 'modelling'), the role of the ontology is to secure
standard and consistent semantics.

> What bad things would happen?

Mostly methodological inconsistency (claim A on the one hand and
implement non A on the other). In some ways, the intended semantics
would become groundless (or more so...).

> Would interoperability in some way be hindered?

Not if BFO was changed, of course. However, if you decide to implement
the choice of allowing process qualities and tie them to processes via
inherence when dealing with your own stuff, obviously you would be
inconsistent with BFO.

PATO allows qualities of processes (I don't know where they falls as a
class) and uses inherence indiscriminately for substantial or
processuals. PATO's notion of inherence is therefore not BFO's (or
RO's for that matter), the former is the more general. Formally, it
means bfo:inheres-in is a subrelation of pato:inheres-in in an OWL
encoding. In a project using both BFO (or rather RO) and PATO, we need
a mapping of this type to make sure we remain consistent with other
OBO ontologies.

> Am I missing something?

No. But I think you are assuming that an ontology is the same as a
vocabulary for a knowledge representation. It's not the case with BFO.
BFO is designed to make salient a number of ontological commitments.
It's not designed to equate with a language. One way of thinking of
BFO in relation to the language needed is similar to the way primitive
stand to defined terms. There's no primitive in BFO's minimal language
for describing processes in terms of properties they have. This is not
entirely the best way of putting it though, but I'm not sure it
matters to further split hairs.

cheers
p

[Pierre Grenon]

On Tue, Oct 25, 2011 at 1:43 PM, Robert Hoehndorf <leec...@leechuck.de> wrote:
>>>>>> "BS" == Barry Smith <phis...@buffalo.edu> writes:
>
>    BS> The argument, all along, is that when we make predications
>    BS> concerning processes we are not referring to some extra entity,
>    BS> as we are when we assert, for example, that a continuant has a
>    BS> suntan or has a disease; rather we are asserting that the
>    BS> process is of a certain type, for instance it is of the type
>    BS> beating process, or of the type beating process with rate 60bpm,
>    BS> and so forth.
>
> The same choice can be made for objects: the apple's red does not
> require an additional entity of the type "red" that inheres in the
> apple, but rather the apple could just be of the type "red" (or "red
> thing"), or it could just be of the type "red, round thing with 10cm
> diameter", etc.

again, its a philosophical argument. In short, apples change but their
fall or their changes of colour don't.

> You have not answered the question about why BFO makes that particular
> choice, rather reiterated the consequences of the choice.

Does it matter?

> And the
> consequences you describe are what makes the use of BFO problematic,
> since they are both counterintuitive and incompatible with the existing
> database/ontology designs.

While I'm now familiar with your view and somewhat sympathetic to it,
or to the frustration expressed with them, I believe, for all it's
worth, that the problem is methodological rather than fundamental, i.e
it is a problem with a certain use of BFO, not with BFO.

p

> R

[Larry Hunter]

On Oct 25, 2011, at 7:50 AM, Barry Smith wrote:

> good enough shape to present at the meeting on November 10-11.

Pardon my ignorance, but what meeting Nov 10-11?

Larry

[Pierre Grenon]

hello,

what's your perception of the adequacy of PATO for the purpose explained below?

Sure, from the BFO standpoint, PATO has fictions and its use of
inherence is wrong (because too generic). But these deviations could
be construed as innocuous syntactic verbosity and some of PATO's terms
even be instrumental in superficial definitions of tentative 'process
types' alluded to by Barry. Neither PATO nor BFO would disagree on the
latter I imagine.

By that light, in PATO, the most valuable things are relations between
processes, which seems to be a large part of what's needed below too.

p

[Cristian Cocos]

On 10/25/2011 8:32, Ludger Jansen wrote:

A pizza with a salty taste is a pizza with a lot of salt in it. However, a pizza with an intensive taste is not a pizza with a lot of intensivity in it!

Surely, nothing prevents you from defining an entity on the basis of the intense taste of the pizza, to the effect that the pizza has an intense taste iff it literally possesses that entity called, for the lack of a better word, intenseness. The situation would be similar to many other physical entities that have been theoretically postulated first, and then "discovered" ostensively (or "directly experienced/isolated") later. At any rate,  this whole debate looks to me to be moot as long as you do not make it clear that what you are actually talking about is asserted/primitive entities as opposed to defined: surely, based on an individual's having a quality one can define a type made up of all of the individuals that have that quality and vice-versa. Whichever comes fist in some ontological order will then come down to one's choice of primitives: what primitives are better choices then others? (Yep, an epistemologist wrinkle, which I'm having a hard time shaking off.)

C

[Ludger Jansen]

Am 25.10.2011 17:42, schrieb Cristian Cocos:

On 10/25/2011 8:32, Ludger Jansen wrote:

A pizza with a salty taste is a pizza with a lot of salt in it. However, a pizza with an intensive taste is not a pizza with a lot of intensivity in it!

Surely, nothing prevents you from defining an entity on the basis of the intense taste of the pizza, to the effect that the pizza has an intense taste iff it literally possesses that entity called, for the lack of a better word, intenseness. The situation would be similar to many other physical entities that have been theoretically postulated first, and then "discovered" ostensively (or "directly experienced/isolated") later.

Science fiction apart, my supermarket sells salt but not intenseness in boxes. The point is that in order to get an intensive taste you take other things like peppers etc. and a lot of them. That the taste is intense is then a result of these spicy toppings.

At any rate,  this whole debate looks to me to be moot as long as you do not make it clear that what you are actually talking about is asserted/primitive entities as opposed to defined: surely, based on an individual's having a quality one can define a type made up of all of the individuals that have that quality and vice-versa. Whichever comes fist in some ontological order will then come down to one's choice of primitives: what primitives are better choices then others? (Yep, an epistemologist wrinkle, which I'm having a hard time shaking off.)

In the pizza analogy, the choice is clear. You cannot buy intenseness, and you cannot replace adding peppers by adding some basic intenseness instead, because this you cannot buy. But I grant that often the case may not be thus clear!

Ludger

C

--
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.

[Jessica Turner]

So with all that, what happened to Melissa's need to identify and compare certain processes' rates? How is that represented?

Thanks,

Jessica

Jessica Turner, Ph.D.
Associate Professor, Translational Neuroscience
Mind Research Network
1101 Yale Blvd. NE
Albuquerque, New Mexico 87106
New Ph: 505-272-2869
email: jtu...@mrn.org

[Chris Mungall]

On Oct 25, 2011, at 6:32 AM, Ludger Jansen wrote:

Robert, maybe an analogy can help:

A pizza with a salty taste is a pizza with a lot of salt in it. However, a pizza with an intensive taste is not a pizza with a lot of intensivity in it!

I'm not sure I've ever eaten an intensive pizza, so I'm not really following

Likewise, an apple needs the property of redness as "an additional entity" to be red -- for the apple itself could also be green, thus as such it does not account for its being red.

But if you have a process, things are different. My going from A to B is a process. What makes this a fast process? Do we have to add some fastness to the process? No, it's simply that I have to go quicker. It's not an additional property belonging to a process, but ME changing in a different way.

I'm happy for physical entities to have the property of speed, this seems in line with accounts from physics.

But the property of duration belongs to the process, and properties derived from this such as rate

[Ludger Jansen]

Am 25.10.2011 18:20, schrieb Chris Mungall:

On Oct 25, 2011, at 6:32 AM, Ludger Jansen wrote:

Robert, maybe an analogy can help:

A pizza with a salty taste is a pizza with a lot of salt in it. However, a pizza with an intensive taste is not a pizza with a lot of intensivity in it!

I'm not sure I've ever eaten an intensive pizza, so I'm not really following

Then, maybe, you would like to bake one ;-)

Likewise, an apple needs the property of redness as "an additional entity" to be red -- for the apple itself could also be green, thus as such it does not account for its being red.

But if you have a process, things are different. My going from A to B is a process. What makes this a fast process? Do we have to add some fastness to the process? No, it's simply that I have to go quicker. It's not an additional property belonging to a process, but ME changing in a different way.

I'm happy for physical entities to have the property of speed, this seems in line with accounts from physics.

That's not exactly what I wanted to suggest.

It might indeed be possible to ascribe speed to physical entities. But you have to go for momentary speed in this case, thus deviating from the standard definition of speed as: Delta s/Delta t -- because any Delta t is stretched out in time and thus speed could not be ascribed at a single point of time, which would be necessary to make it a property, i.e. a dependent continuant. The mathematical bugfix here is, of course, a lim Delta t --> 0, but would that suffice ontologically?

My suggestion rather is: I am a participant in a certain process that has a certain outcome after a certain amount of time -- and this outcome can be reached faster or slower than in other cases (or than in the canonical case).

Hence, from the BFO point of view speed is very complex: You need a span ontology and several snap ontologies (at different time points) in order to model it. (Or can we get that cheaper?)

 

But the property of duration belongs to the process, and properties derived from this such as rate

In any case, is there a reason why you want to distinguish speed and rate? From me they seem to be on the same footing, ontologically. If a body had a speed as its property, why not having development rates as properties of brains?

Best
Ludger

[Dr. Adam M. Goldstein PhD MSLIS]

I take it that Barry's view is that it would misrepresent reality and that if it's made clear enough how to handle the various cases, the ontologies using whatever the recommended rules are will be interoperable.

This is a speculation however.

------------------
Adam M. Goldstein PhD, MSLIS
--
z_calif...@shiftingbalance.org
http://www.shiftingbalance.org
http://www.twitter.com/shiftingbalance
--
http://www.itis.gov/servlet/SingleRpt/SingleRpt?search_topic=TSN&search_value=180621
--
Associate Editor
Reviews Editor
Evolution: Education & Outreach
http://www.springer.com/life+sci/journal/12052
--
Spellman 205
(914) 637-2717 (msg)
--
Dept of Philosophy
Iona College
715 North Avenue
New Rochelle NY 10801
http://www.iona.edu/faculty/agoldstein

[Wacek Kusnierczyk]

On Mon, 24 Oct 2011 13:49:00 -0700, Chris Mungall <cjmu...@lbl.gov>
wrote:

Let's say you'd abandon BFO and develop ontologies based on what's
appropriate for your pragmatic needs rather than speculations about
qualities. What bad things would happen?

vQ

[Chris Mungall]

Rhetorical question?

For one thing it would create a schism rendering it harder to interoperate with ontologies that do use BFO, which most of us I think would count as a Bad Thing.

I believe in moving forward based on consensus, rather than splintering and forming factions, based on one person's needs. I'd hope that if there is a consensus amongst users that some part of BFO needs fixing then it would be fixed - this is no different from any other ontology.

> vQ

[Alexander Garcia]

This is Interesting:

> For one thing it would create a schism rendering it harder to interoperate with ontologies that do use BFO, which most of us I think would count as a Bad Thing.

Most ontologies do not make any use of BFO, so far so good. Even in the bio domain, there Is a good number of ontologies not making use of BFO. I know that I will get a lot of heat because of these words, but come on most ontologies dont make any use of upper level ontologies. The proof is in the pooding, using BFO as an interoperability strategy, or any upper level ontology for that matter, will only be tested once u have a lot of ontos following the same strategy, and following it in the same way -even if u use BFO, some one else also making use of it may interpret BFO differently.

Sent from my iPad

[Pierre Grenon]

Hi Melissa,

would you have the time to explain these four particular examples or
at least to criticise the tentative understanding below? You don't
have to explain the minute biology (it will be mostly lost on me I'm
afraid), but I have a hard time making sure I understand what the
examples illustrate in terms of i) what is said and, in the first
place, ii) about what it is said. I'll ask questions below.

On Tue, Oct 25, 2011 at 2:50 PM, Melissa Haendel <hae...@ohsu.edu> wrote:

> I need to be able to represent that the rate of a process in one organism is
> decreased relative to the rate of the process in another organism.
> eg. brain development is occurring at a reduced rate in a shh -/- zebrafish
> mutant, relative to a wildtype sibling of the AB strain.

do we speak of two specific instances (of two different types of
organisms, respectively) or of two different types of organisms (but
not about particular instances)? That is, is it i) the particular
instance of the brain development of, say, Mut1 which is compared to
the particular instance of development of a particular instance of the
wildtype, say Wild1, or is it any brain development in some mutant of
a given kind that is compared to any brain development of fishes of a
wildtype?

At first glance it seems that all that is needed for representing the
claim are the following elements:
- term for type of organism 1
- term for type of organism 2
- a comparative relation that could read: has higher rate of brain
develeopment than

Would the requirred representation fit the bill if it were a sentence
about the types of organism (or about some instances thereof). Or is
the representation one that takes brain development (as processes) as
the subjects of comparison?

In the latter case, it seems sufficient to have:
- a term for the type of processes which is brain development
(accessorily:
term for type of organism 1
term for type of organism 2
a relation of participation of sort to characterise brain developments
in instances of a type or of another)
- a comparative relation between brain developments processes that
could read: has higher rate than

While I'm not sure what the representation target, I don't see yet a
need for rates as entities in their own rights.

> I also need to be able to represent that a process has a change in onset
> relative to another process. Some examples are:
> chondrogenesis has a late onset relative to normal (relative to the
> development of the organism as a whole)
> chondrogenesis has a late onset relative to osteogenesis (one processs is
> delayed in its start relative to another)

Could you please explain what the following mean:
2) 'X has late onset relative to normal (relative to the development
of the organism as a whole)'
3) 'X has late onset relative to Y'

I read 2) as meaning something like 'is a late part of the development
of an organism'.

I read 3) as meaning 'X starts after the start of Y'.

3) can be handled with simple definitional extensions in BFO. 2) is
slightly more unusual.

> I need to be able to describe a process occurring asynchronously.
> for example, asynchronous pigmentation.

Let's call that 4). Could you please explain what's at stake here?
what is an asynchronous process in this context?

> Scientists need to be able to record these aspects of processes as part of
> phenotype annotation. We don't care what we call them,

That's good!

> we just need to be
> able compute on them. For example, find me all organisms that have
> asynchrony in all neural crest development processes.

ok. There's a number of stategies here. Having an asyncronous 'process
quality' is not better motivated by such use case than having an
'asynchronous neural crest development' type.

> I hope that some of these examples (and the many others that I am sure you
> have considered) are helpful in making BFO work for this purpose. Scientists
> need a straightforward mechanism to represent process phenotypes, and
> ontologists need a way to compute on them. We need to ensure that whatever
> is done in BFO to this end isn't so complex so as to deter the scientists
> from recording these very important aspects of phenotype or we will not get
> the data that we need to do our informatics work.

I hope scientists never have to worry about the level at which BFO operates.

cheers
p

[Adam M. Goldstein]

On Oct 26, 2011, at 1:40, Chris Mungall <cjmu...@lbl.gov> wrote:

>
> On Oct 25, 2011, at 9:25 PM, Wacek Kusnierczyk wrote:
>
>> On Mon, 24 Oct 2011 13:49:00 -0700, Chris Mungall <cjmu...@lbl.gov>
>> wrote:
>>> Can someone remind me why qualities of processes (e.g. rates) are
>>> disallowed? Is there documentation e.g. on a wiki somewhere?
>>>
>>> Let's just say that BFO were to allow qualities to inhere in
>>> processes. There would be many pragmatic advantages. Modeling would be
>>> easier and more in line with scientists' intuitions.
>>>
>>> What bad things would happen? Would interoperability in some way be
>>> hindered? Am I missing something?
>>
>> Let's say you'd abandon BFO and develop ontologies based on what's
>> appropriate for your pragmatic needs rather than speculations about
>> qualities. What bad things would happen?
>
> Rhetorical question?
>

For many people I think, no.

> For one thing it would create a schism rendering it harder to interoperate with ontologies that do use BFO, which most of us I think would count as a Bad Thing.
>

If the question is, what benefits does using the BFO bring, and the answer is, there is greater interoperability with other ontologies that use the BFO, then, the question still remains, why use the BFO? Rather than an in-principle promise that interoperability will increase, are there any measurements we can make that will show this? The BFO strategy is not the only one for interoperability; and there are values that can override interoperability.

[Wacek Kusnierczyk]

On Tue, 25 Oct 2011 22:40:42 -0700, Chris Mungall <cjmu...@lbl.gov>

wrote:
> On Oct 25, 2011, at 9:25 PM, Wacek Kusnierczyk wrote:
>
>> On Mon, 24 Oct 2011 13:49:00 -0700, Chris Mungall <cjmu...@lbl.gov>
>> wrote:
>>> Can someone remind me why qualities of processes (e.g. rates) are
>>> disallowed? Is there documentation e.g. on a wiki somewhere?
>>>
>>> Let's just say that BFO were to allow qualities to inhere in
>>> processes. There would be many pragmatic advantages. Modeling would be
>>> easier and more in line with scientists' intuitions.
>>>
>>> What bad things would happen? Would interoperability in some way be
>>> hindered? Am I missing something?
>>
>> Let's say you'd abandon BFO and develop ontologies based on what's
>> appropriate for your pragmatic needs rather than speculations about
>> qualities. What bad things would happen?
>
> Rhetorical question?

wasn't meant to be.

is bfo the only way to go? it's understandable that using a common
framework or standard may be considered a step towards better
interoperability. it's somewhat less obvious (as witnessed, e.g., on
the ontolog forum) that choosing one specific top-level highly abstract
ontology is the silver bullet. it's even less obvious that bfo is the
one to choose.

>
> For one thing it would create a schism rendering it harder to
> interoperate with ontologies that do use BFO, which most of us I think
> would count as a Bad Thing.

ouch, SCARY. so modifying bfo according to some of the arguments made
on this list would obviously keep all those bfo-compliant ontologies
interoperable just in virtue of the fact that they all follow bfo?

> I believe in moving forward based on consensus, rather than
> splintering and forming factions, based on one person's needs. I'd
> hope that if there is a consensus amongst users that some part of BFO
> needs fixing then it would be fixed - this is no different from any
> other ontology.

you can splinter, form a faction, develop a better ontology and show it
as a proof of concept. if you succeed, others will likely follow. if
you don't, you'll still provide a useful piece of evidence. that's
research. it's been repeatedly announced that obo foundry and bfo are
research activities. well, research is about exploration, not rigidly
sticking with one particular framework. think different.

vQ

[Wacek Kusnierczyk]

On Wed, 26 Oct 2011 10:34:31 +0200, Alexander Garcia
<alexg...@gmail.com> wrote:
> This is Interesting:
>
>> For one thing it would create a schism rendering it harder to interoperate with ontologies that do use BFO, which most of us I think would count as a Bad Thing.
>
> Most ontologies do not make any use of BFO, so far so good. Even in
> the bio domain, there Is a good number of ontologies not making use of
> BFO. I know that I will get a lot of heat because of these words, but
> come on most ontologies dont make any use of upper level ontologies.
> The proof is in the pooding, using BFO as an interoperability
> strategy, or any upper level ontology for that matter, will only be
> tested once u have a lot of ontos following the same strategy, and
> following it in the same way -even if u use BFO, some one else also
> making use of it may interpret BFO differently.

... which is very likely if you do not have a philosopher on the team,
and certain if you do.

vQ

[Michel Dumontier]

Most ontologies that purport to follow BFO, may do so in spirit of a unifying framework (OBO Foundry), but clearly are confused by / do not conform to BFO/RO *intended* semantics. 

 

> I believe in moving forward based on consensus, rather than
> splintering and forming factions, based on one person's needs. I'd
> hope that if there is a consensus amongst users that some part of BFO
> needs fixing then it would be fixed - this is no different from any
> other ontology.

you can splinter, form a faction, develop a better ontology and show it
as a proof of concept.  if you succeed, others will likely follow.  if
you don't, you'll still provide a useful piece of evidence.  that's
research.  it's been repeatedly announced that obo foundry and bfo are
research activities.  well, research is about exploration, not rigidly
sticking with one particular framework.  think different.

+1

m. 

[Robert Hoehndorf]

>>>>> "LJ" == Ludger Jansen <ludger...@uni-rostock.de> writes:

Hi,

LJ> Likewise, an apple needs the property of redness as "an
LJ> additional entity" to be red -- for the apple itself could also
LJ> be green, thus as such it does not account for its being red.
LJ> But if you have a process, things are different. My going from A
LJ> to B is a process. What makes this a fast process? Do we have to
LJ> add some fastness to the process? No, it's simply that I have
LJ> to go quicker. It's not an additional property belonging to a
LJ> process, but ME changing in a different way.

If an apple can have a "redness" at a time point, why can a process not
have a "duration"? The need for properties' changing over time seems to
be not relevant to the question whether processes can have qualities or
not.

R

[Robert Hoehndorf]

>>>>> "CM" == Chris Mungall <cjmu...@lbl.gov> writes:

CM> For one thing it would create a schism rendering it harder to
CM> interoperate with ontologies that do use BFO, which most of us I
CM> think would count as a Bad Thing.

Can you point me to one case where the application of BFO has lead to
"interoperability" between ontologies that would not have been possible
without it? I am quite interested in what kind of "operations" were
enabled through BFO's use.

CM> I believe in moving forward based on consensus, rather than
CM> splintering and forming factions, based on one person's
CM> needs. I'd hope that if there is a consensus amongst users that
CM> some part of BFO needs fixing then it would be fixed - this is
CM> no different from any other ontology.

This seems to be the core of the debate: will biomedical resources,
annotation guidelines, ontologies, etc. be made to change so that they
accommodate the view on the world imposed by one philosophical theory
called BFO, or should there be an evidence-driven, empirical way to
connect the existing resources and make the relevant distinctions
explicit? Sometimes, it seems to me that BFO and parts of OBO seem to be
taken the first route: make a philosophically-motivated decision, and
convince people to change whatever they are doing to accommodate it. If
this will be the most successful strategy is still an open question, I
think.

Rob.

[Robert Hoehndorf]

>>>>> "PG" == Pierre Grenon <pierre...@gmail.com> writes:

>> Let's just say that BFO were to allow qualities to inhere in
>> processes.

PG> BFO would then need to account for the theory of such
PG> 'inherence'. There is something misleading in the
PG> assumption. While inherence already has a precise technical
PG> meaning in BFO, the assumption seems to be that inherence is the
PG> relation that would obtain between process attributes/qualities
PG> and processes. This assumption seems to be relying on the
PG> superficial linguistic features, predication in English, in
PG> particular. These are not necessarily to be brushed aside but
PG> they are, methodologically speaking, without compelling
PG> consequences for the ontology.

As you say, "philosophical doctrine" is the main methodology that was
followed when constructing the BFO. Where does this doctrine come from,
and what is its purpose? If it is intended to serve science and
scientific data integration, perhaps it can be changed (although that
may go against the "doctrine" part).

>> There would be many pragmatic advantages.

PG> I think the advantages you are expecting are tied to having a
PG> knowledge representation with an expressive vocabulary. This is
PG> arguably not the same thing as having an ontology.

I see. So are you saying that BFO is a philosophical theory that is not
intended to be useful for data integration in science, and instead a
yet-to-be-developed "knowledge representation" artifact will be what can
eventually be applied to serve science?

Rob.

[Wacek Kusnierczyk]

now here's a rhetorical question: do you really think that convincing
biologists to follow a specific (and seemingly controversial and
obstructing) philosophical doctrine might be the most successful
strategy?

vQ

[Melissa Haendel]

Hi Pierre,

absolutely, thanks for asking.

On Oct 26, 2011, at 3:22 AM, Pierre Grenon wrote:

Hi Melissa,

would you have the time to explain these four particular examples or
at least to criticise the tentative understanding below? You don't
have to explain the minute biology (it will be mostly lost on me I'm
afraid), but I have a hard time making sure I understand what the
examples illustrate in terms of i) what is said and, in the first
place, ii) about what it is said. I'll ask questions below.

On Tue, Oct 25, 2011 at 2:50 PM, Melissa Haendel <hae...@ohsu.edu> wrote:

I need to be able to represent that the rate of a process in one organism is

decreased relative to the rate of the process in another organism.

eg. brain development is occurring at a reduced rate in a shh -/- zebrafish

mutant, relative to a wildtype sibling of the AB strain.

do we speak of two specific instances (of two different types of
organisms, respectively) or of two different types of organisms (but
not about particular instances)? That is, is it i) the particular
instance of the brain development of, say, Mut1 which is compared to
the particular instance of development of a particular instance of the
wildtype, say Wild1, or is it any brain development in some mutant of
a given kind that is compared to any brain development of fishes of a
wildtype?

Very good question. This may seem trivial, but actually from the perspective of annotation it isn't. There is a longstanding discussion about whether species/strains/genotypes are instances or types. From the annotation perspective here is what happens: 

A researcher has two dishes of fishes (lets call them groups of instances) of organism type 1 (mut1) and type 2 (Wild1). In one or more of the fishes of type mut1 in dish one, the brain development is occurring at a higher rate than those instances of Wild1 fishes in the other dish. The researcher, depending on if it meets some statistical standard (or not..) will then record that brain development in organism mut1 has a higher rate than that of organism Wild1. They would infer that their observation applies to all organisms of mut1 relative to Wild1, however this information would be attributed ideally with the person, date, experimental conditions etc. This annotation may conflict with other researcher's observations on a different collection of mut1 and Wild1 organism instances.

At first glance it seems that all that is needed for representing the
claim are the following elements:
- term for type of organism 1
- term for type of organism 2
- a comparative relation that could read: has higher rate of brain
develeopment than

Would the requirred representation fit the bill if it were a sentence
about the types of organism (or about some instances thereof). Or is
the representation one that takes brain development (as processes) as
the subjects of comparison?

In the latter case, it seems sufficient to have:
- a term for the type of processes which is brain development
(accessorily:
term for type of organism 1
term for type of organism 2
a relation of participation of sort to characterise brain developments
in instances of a type or of another)
- a comparative relation between brain developments processes that
could read: has higher rate than

While I'm not sure what the representation target, I don't see yet a
need for rates as entities in their own rights.

I think the simple need pointed out in #1 suffices, as long as I can create the composed classes to create the annotation. For example, I need the GO class 'brain development', I need the genotypes or organism types, and then *something* to make the comparative statement about the rate. The annotator isn't likely recording anything about participants in this case, although by definition, the GO class 'brain development' would imply that the brain is a participant in this process. If the brain is also smaller, spongier, smaller mass, etc, we'd record that as well, as a quality of an anatomical entity. However, in other cases we do have a need to record a difference in participants in a process, for example if in one species a given anatomical part participates in some induction process, whereas in another it doesn't.

I also need to be able to represent that a process has a change in onset

relative to another process. Some examples are:

chondrogenesis has a late onset relative to normal (relative to the

development of the organism as a whole)

chondrogenesis has a late onset relative to osteogenesis (one processs is

delayed in its start relative to another)

Could you please explain what the following mean:
2) 'X has late onset relative to normal (relative to the development
of the organism as a whole)'
3) 'X has late onset relative to Y'

I read 2) as meaning something like 'is a late part of the development
of an organism'.

I read 3) as meaning 'X starts after the start of Y'.

3) can be handled with simple definitional extensions in BFO. 2) is
slightly more unusual.

ok so 2 says that dish1 of organisms (lets say mut2 type) has a later start time of chondrogenesis than organisms in dish2 (say Wild2).

This assumes normal developmental time, regular conditions. development is always a bell curve even in these conditions. Some fishes are smaller, slower, so we measure relative to the whole fish. If chondrogenesis starts later in any given fish, but the whole fish is delayed, it isn't signficant. Its when the rest of the fish is developing on schedule and chondrogenesis is delayed in its beginning that would be significant.

for 3, its really a more specific case of 2, but now I am comparing not process a (chondrogenesis) to process b (development of the whole organism) but rather the start times of two different processes process a (chondrogenesis) to process c (osteogenesis). The problem is in cases like 2, the annotator won't record anything other than process a has a late onset- its implied that it is relative to process b (development of the whole organism). 

I need to be able to describe a process occurring asynchronously.

for example, asynchronous pigmentation.

Let's call that 4). Could you please explain what's at stake here?
what is an asynchronous process in this context?

So normally this particular process occurs synchronously in specific locations throughout the organism. If the pigmented cells don't become pigmented in their normal synchronous manner, the annotator would record that the process is occurring asynchronously. 

Scientists need to be able to record these aspects of processes as part of

phenotype annotation. We don't care what we call them,

That's good!

:-) 

we just need to be

able compute on them. For example, find me all organisms that have

asynchrony in all neural crest development processes.

ok. There's a number of stategies here. Having an asyncronous 'process
quality' is not better motivated by such use case than having an
'asynchronous neural crest development' type.

an annotator wouldn't see the difference here. They have to find a process class (GO pigmentation) and be able to state something about how the process is not synchronized across the organism. GO will not contain 'asynchronous pigmentation', it needs to be post-composed. And the annotation is referencing something non-canonical - this gets into how we use these canonical ontologies for annotating non-canonical observations. If all 'pigmentation' was defined as being synchronous (which it isn't - not a good example for this) then this annotation would be violate the definition. But this is a whole other can of worms and is out of scope for this purpose.

I hope that some of these examples (and the many others that I am sure you

have considered) are helpful in making BFO work for this purpose. Scientists

need a straightforward mechanism to represent process phenotypes, and

ontologists need a way to compute on them. We need to ensure that whatever

is done in BFO to this end isn't so complex so as to deter the scientists

from recording these very important aspects of phenotype or we will not get

the data that we need to do our informatics work.

I hope scientists never have to worry about the level at which BFO operates.

Unfortunately we care very much, as what BFO does directly affects the ease at which we have the ability to use ontologies properly for annotation, and the ability of informaticists to mine these annotations leveraging the inferencing of the ontologies used for this purpose. If BFO becomes overly complex to the point where scientists cannot use it directly, we will need to use shortcut relations and the like or there will be no interoperability between ontologies, annotation data sets, or consistency of ontology use. 

I hope this is helpful.

Best,

Melissa

[Melissa Haendel]

See my other response ;-).

--
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.

[Albert Goldfain]

Suppose there were a version of BFO (call it BFO+) that contained the type 'dependent occurrent' as a subtype of occurrent, and was defined in such a way so that 'rate' is a dependent occurrent in BFO+.

Further, suppose that in the current BFO, we make 'rate' a subtype of Entity.

I would submit that biologists will be looking for terms like 'rate' from which to subtype (e.g., 'respiration rate'), and they will not be looking for 'dependent occurrent'.  As a result, they will develop fairly stable hierarchies (e.g., human respiration rate is a subtype of respiration rate) and BFO can do the philosophy under the hood.

AG
 

[Stefan Schulz]

>
> If an apple can have a "redness" at a time point, why can a process not
> have a "duration"?

I think there is a consensus that all processes have a duration.
Whether we name the duration of a process a "process quality", and the
relation between the duration and the process "inherence", is just a
labelling issue and therefore a matter of convention. There may be
good reasons to
be more restrictive with the scope of relations in order to avoid
disjunctions in domain and range restrictions. I can therefore live
with a convention that restricts the range of "inheres_in" to
independent continuants.

What is still open for me:

1.) How to relate processes with their durations, e.g. by a relation
"has duration" ?

2.) Whether the alleged process qualities can be expressed by
combining the duration of a process with the qualities of its
participants.

3.) Given 2.) is true, whether process qualities such as in phenotype
ontologies can be allowed as a kind of "second class entities", for
the sake of simplicity. I am sure that if BFO disallows process
qualities, BFO users won't care and introduce whatsower relation in
order to, e.g. , connect " course of disease" with "chronic".

Stefan

--

Stefan SCHULZ  (Univ.-Prof. Dr. med.)

Institut für Medizinische Informatik,
Statistik und Dokumentation
Medizinische Universität Graz
Auenbruggerplatz 2/V
8036 Graz (Austria)

http://www.medunigraz.at/imi
 +43 (0)316 385 16939
 +43 (0)316 385 13201

 http://purl.org/steschu
 mailto:stefan...@medunigraz.at
Skype: stschulz

[   home: Afritschgasse 32/3
[   8020 Graz (Austria)
[   mobile:  +43 (0)699 150 96270

[Melissa Haendel]

Hi Albert,

just one comment..

On Oct 26, 2011, at 8:14 AM, Albert Goldfain wrote:

On Wed, Oct 26, 2011 at 10:52 AM, Wacek Kusnierczyk <wa...@idi.ntnu.no> wrote:

On Wed, 26 Oct 2011 14:27:17 +0100, Robert Hoehndorf
<leec...@leechuck.de> wrote:
>>>>>> "CM" == Chris Mungall <cjmu...@lbl.gov> writes:
>

> This seems to be the core of the debate: will biomedical resources,
> annotation guidelines, ontologies, etc. be made to change so that they
> accommodate the view on the world imposed by one philosophical theory
> called BFO, or should there be an evidence-driven, empirical way to
> connect the existing resources and make the relevant distinctions
> explicit? Sometimes, it seems to me that BFO and parts of OBO seem to be
> taken the first route: make a philosophically-motivated decision, and
> convince people to change whatever they are doing to accommodate it. If
> this will be the most successful strategy is still an open question, I
> think.

now here's a rhetorical question: do you really think that convincing
biologists to follow a specific (and seemingly controversial and
obstructing) philosophical doctrine might be the most successful
strategy?

Suppose there were a version of BFO (call it BFO+) that contained the type 'dependent occurrent' as a subtype of occurrent, and was defined in such a way so that 'rate' is a dependent occurrent in BFO+.

Further, suppose that in the current BFO, we make 'rate' a subtype of Entity.

I would submit that biologists will be looking for terms like 'rate' from which to subtype (e.g., 'respiration rate'), and they will not be looking for 'dependent occurrent'. 

I really wouldn't worry about this too much. All applications using BFO ontologies already have the need to hide these upper level classes. Those of us who know BFO/obo Foundry will know where to look.  The need here is to have the ability to properly logically post-compose things like 'human respiration rate' from 'human', 'respiration' and 'rate'. As you say, this is what BFO needs to do under the hood- provide the proper relations and types of these entities. 

As a result, they will develop fairly stable hierarchies (e.g., human respiration rate is a subtype of respiration rate) and BFO can do the philosophy under the hood.

AG
 

vQ

--
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.

--
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.

[Wacek Kusnierczyk]

On Wed, 26 Oct 2011 11:14:37 -0400, Albert Goldfain
<albertg...@gmail.com> wrote:
>
> Suppose there were a version of BFO (call it BFO+) that contained the
> type 'dependent occurrent' as a subtype of occurrent, and was defined
> in such a way so that 'rate' is a dependent occurrent in BFO+.
>
> Further, suppose that in the current BFO, we make 'rate' a subtype of
> Entity.
>
> I would submit that biologists will be looking for terms like 'rate'
> from which to subtype (e.g., 'respiration rate'), and they will not be
> looking for 'dependent occurrent'.  As a result, they will develop
> fairly stable hierarchies (e.g., human respiration rate is a subtype
> of respiration rate) and BFO can do the philosophy under the hood.
>

i like the idea: let the biologists do their job as they seem
appropriate, practical, and useful, and the bfo folks do theirs behind
the scenes so that no one has to know it even exists.

vQ

[Pierre Grenon]

On Wed, Oct 26, 2011 at 2:35 PM, Robert Hoehndorf <leec...@leechuck.de> wrote:
>>>>>> "PG" == Pierre Grenon <pierre...@gmail.com> writes:
>
>    >> Let's just say that BFO were to allow qualities to inhere in
>    >> processes.
>    PG> BFO would then need to account for the theory of such
>    PG> 'inherence'.  There is something misleading in the
>    PG> assumption. While inherence already has a precise technical
>    PG> meaning in BFO, the assumption seems to be that inherence is the
>    PG> relation that would obtain between process attributes/qualities
>    PG> and processes. This assumption seems to be relying on the
>    PG> superficial linguistic features, predication in English, in
>    PG> particular. These are not necessarily to be brushed aside but
>    PG> they are, methodologically speaking, without compelling
>    PG> consequences for the ontology.
>
> As you say, "philosophical doctrine" is the main methodology that was
> followed when constructing the BFO. Where does this doctrine come from,
> and what is its purpose?

No, not doctrine as a methodology. Philosophical investigations, of
some stripe, as a methodology. This is opposed to, for example, corpus
based analysis methodology that will create semantic networks of terms
used as an ontology. This is also opposed to random assembly of
whatever somebody finds useful or more telling as a term.

Whichever the methodology, however, the work results in some artifact.
In the case of BFO, the work results in the formulation of a theory.
This, in itself, could be appealing, but it doesn't have to. However,
when we refer to the theory, it needs to be understood as a doctrine
so that we can agree about what we are talking about... If we argue
about BFO, we are not arguing about what BFO could be.

This is merely a requirement for any rational debate.

> If it is intended to serve science and
> scientific data integration, perhaps it can be changed (although that
> may go against the "doctrine" part).

It is intended as a basic formal ontology of reality. The assumption
in speculating about its usefulness for any type of purpose is that
when used in combination with something which too is about reality, we
are at least more or less aiming in the right direction.

That it can be changed is also part of the methodological tenets of
BFO. It can't be changed, however, just because somebody doesn't like
it or see no use in it at some point or another.

>    >> There would be many pragmatic advantages.
>    PG> I think the advantages you are expecting are tied to having a
>    PG> knowledge representation with an expressive vocabulary. This is
>    PG> arguably not the same thing as having an ontology.
>
> I see. So are you saying that BFO is a philosophical theory that is not
> intended to be useful for data integration in science, and instead a

I'm not sure how your can read this out of what I wrote.

I'm saying that BFO is an ontology, not a logical calculus. The
vocabulary you would use to specify BFO would be a subset of the
vocabulary you could use for knowledge representation based on BFO.

I'm not saying, but I'm assuming that BFO is to be understood as a
formal ontology, not an ontology of any domain in particular. In that
sense it does not by itself satisfy specific domain ontological needs
(you can't read out of BFO that there are fishes). This lends credit
to the view that the specification vocabulary for BFO is inherently
limited. Aside from domain specific additions (fishes), it's still
unclear whether a useful vocabulary for knowledge representation based
on BFO can be a mere definitional extension. It's also unclear that it
should be.

> yet-to-be-developed "knowledge representation" artifact will be what can
> eventually be applied to serve science?

BFO's a tool of a very generic sort. Indeed, there's much to do if you
want to apply it to serve any purpose. Is there something wrong with
this?

Hope this makes some sense, I wrote this in a very noisy room ;)
p

> Rob.

[Jessica Turner]

Wait, if as Pierre suggests below, there could be a relationship between processes that captures the idea that one process has a faster rate than another process, and yet processes can't have qualities like rates, how would that work?  What should you be able to infer from knowing that that relationship holds between the two processes, if it isn't the naive idea that the processes each have a rate and one of them is greater or faster than the other?

[Pierre Grenon]

If you're confident you can evaluate your pragmatic needs without
speculating about qualities, maybe the BFO methodology doesn't
coinincide with yours. But I'm not sure how you evaluate the fitness
of an ontology to your pragmatic needs then. Obviously, it depends on
what you're trying to do etc, which is hard to discuss in the
abtsract...

> What bad things would happen?

Assume one have in mind a motivation for ontology that more or less
squares with a well delineated purpose in a more or less well
delineated domain. The worst thing that could happen in my view would
be to be thrown even deeper into the plaguing cycle of reinventing the
wheel and in short doing everything but what is fundamentally of
practical interests, that is, an adequate domain ontology.

Concretely, assume we pick a domain and do things from scratch. Having
such license would be very exciting. But quickly would arise the need
for some tool comparable to BFO. BFO's a tool for constructing more
tools.

p

>
> vQ

[Chris Mungall]

On Oct 26, 2011, at 7:52 AM, Wacek Kusnierczyk wrote:

Well when you put it like that, no.

I'd hope there to be some leeway - such as sticking to its realist principles but deferring to pragmatism in those cases where there is no obvious and non-controversial appeal to realism. My intuition is that biologists could get behind that.

[Chris Mungall]

On Oct 26, 2011, at 6:27 AM, Robert Hoehndorf wrote:

>>>>>> "CM" == Chris Mungall <cjmu...@lbl.gov> writes:
>
> CM> For one thing it would create a schism rendering it harder to
> CM> interoperate with ontologies that do use BFO, which most of us I
> CM> think would count as a Bad Thing.
>
> Can you point me to one case where the application of BFO has lead to
> "interoperability" between ontologies that would not have been possible
> without it? I am quite interested in what kind of "operations" were
> enabled through BFO's use.

Just time for a quick response here - you, Wacek and Alexander have brought up the same point which I think was implicit in Wacek's original question.

Note I didn't make any claims about increases in interoperability deriving from BFO - it may be the case that any two upper ontologies deliver negligible benefits in isolation but a net detrimental effect on interoperability when used together. This may be more socio-technological than purely technical - e.g. confusion arising from the doubling of the number of abstruse terms, confusion as to which to use when, opposing religious factions warring over upper levels, etc.

>
> CM> I believe in moving forward based on consensus, rather than
> CM> splintering and forming factions, based on one person's
> CM> needs. I'd hope that if there is a consensus amongst users that
> CM> some part of BFO needs fixing then it would be fixed - this is
> CM> no different from any other ontology.
>
> This seems to be the core of the debate: will biomedical resources,
> annotation guidelines, ontologies, etc. be made to change so that they
> accommodate the view on the world imposed by one philosophical theory
> called BFO, or should there be an evidence-driven, empirical way to
> connect the existing resources and make the relevant distinctions
> explicit? Sometimes, it seems to me that BFO and parts of OBO seem to be
> taken the first route: make a philosophically-motivated decision, and
> convince people to change whatever they are doing to accommodate it. If
> this will be the most successful strategy is still an open question, I
> think.

I think this is a reasonable point.

I would like to see an emphasis on evidence-based and empirical approaches over doctrinal approaches.

> Rob.

[Ludger Jansen]

See inline.

Am 26.10.2011 17:17, schrieb Stefan Schulz:
>> If an apple can have a "redness" at a time point, why can a process not
>> have a "duration"?
> I think there is a consensus that all processes have a duration.
> Whether we name the duration of a process a "process quality", and the
> relation between the duration and the process "inherence", is just a
> labelling issue and therefore a matter of convention. There may be
> good reasons to
> be more restrictive with the scope of relations in order to avoid
> disjunctions in domain and range restrictions. I can therefore live
> with a convention that restricts the range of "inheres_in" to
> independent continuants.

It's perfectly okay to say that processes "have" durations. There is
nothing wrong with this English sentence. But durations are no dependent
continuants, and a fortiori no qualities. They are occurrents, just like
processes. Thus the signature of the relation expressed by the "have" in
scare quotes above is <occurrent, occurrent>. In this it is unlike, say,
"inheres-in", which has the signature <dependent continuant, independent
continuant>.

To some extend this could indeed be seen as a labelling issue. But there
is a trade-off between flexible syntax and semantic rigidity and
reasoning strenght.

> What is still open for me:
>
> 1.) How to relate processes with their durations, e.g. by a relation
> "has duration" ?

Here we could help ourselves by introducing a relation like
"has-duration". But velocity and rate cannot be dealt with in the same
way, for the participant of a process can change with different
velocities at different stages of this process.

> 2.) Whether the alleged process qualities can be expressed by
> combining the duration of a process with the qualities of its
> participants.

What is common to the alleged process qualities? It is, me thinks, the
grammtical feature that predicates used to express the alleged process
qualities can be predicated of grammatical subjects that are names of
processes. But this does not imply that they are all of equal standing
ontologically. See the difference between duration and velocity above.

> 3.) Given 2.) is true, whether process qualities such as in phenotype
> ontologies can be allowed as a kind of "second class entities", for
> the sake of simplicity. I am sure that if BFO disallows process
> qualities, BFO users won't care and introduce whatsower relation in
> order to, e.g. , connect " course of disease" with "chronic".

There is a problem indeed if this connection comes along as a wrong
assertion in the ontology. E.g., "Chronicity inheres_in Course of
disease" would not be BFO-compliant. But I see no problem with this
"connection", as long as it is only a term-building activity. E.g., we
could introduce the naming convention "If X is a term for a course of
disease (in which the term "chronic" does not occurr, of course), then
"Chronic_X" is also a term for a course of disease. Problem: How to
minimize multiple inheritence when using this technique?

Best
Ludger

> Stefan
>
>
>
>
>
>
>
>
>

--

PD Dr. Ludger Jansen
Institut f�r Philosophie
Universit�t Rostock
18051 Rostock

[Chris Mungall]

I support this kind of research and experimentation. I don't think it's best conducted by splintering off one set of production ontologies.

It would be better to 'clone' the set of existing ontologies and experiment with swapping in and out different upper levels (or no upper level) and testing against a set of competency questions. I'm surprised no one has done this already in fact (or perhaps they have and I'm just unaware).

[Adam M. Goldstein]

On Oct 26, 2011, at 12:48, Chris Mungall <cjmu...@lbl.gov> wrote:

>
> On Oct 26, 2011, at 7:52 AM, Wacek Kusnierczyk wrote:
>
>> On Wed, 26 Oct 2011 14:27:17 +0100, Robert Hoehndorf
>> <leec...@leechuck.de> wrote:
>>>>>>>> "CM" == Chris Mungall <cjmu...@lbl.gov> writes:
>>>
>>> This seems to be the core of the debate: will biomedical resources,
>>> annotation guidelines, ontologies, etc. be made to change so that they
>>> accommodate the view on the world imposed by one philosophical theory
>>> called BFO, or should there be an evidence-driven, empirical way to
>>> connect the existing resources and make the relevant distinctions
>>> explicit? Sometimes, it seems to me that BFO and parts of OBO seem to be
>>> taken the first route: make a philosophically-motivated decision, and
>>> convince people to change whatever they are doing to accommodate it. If
>>> this will be the most successful strategy is still an open question, I
>>> think.
>>
>> now here's a rhetorical question: do you really think that convincing
>> biologists to follow a specific (and seemingly controversial and
>> obstructing) philosophical doctrine might be the most successful
>> strategy?
>
> Well when you put it like that, no.
>
> I'd hope there to be some leeway - such as sticking to its realist principles but deferring to pragmatism in those cases where there is no obvious and non-controversial appeal to realism. My intuition is that biologists could get behind that.
>

I don't think the instantiation principle (which is really the basis of BFO "realism" is) is a problem for people. [1] The difficulty rather arises when someone wants to represent an entity or relationship that doesn't seem to fit into the BFO. Now the task is to reconstruct it in BFO terms. Here it is difficult for someone not a BFO specialist to know what to do. If there were a way to quickly assess whether there is a sensible way to do the reconstruction, and then if not, go the pragmatic route, I think that many people would be happier with the BFO. Some people's particular needs might be met right away in BFO terms by a BFO expert.

Adam

[1] For some uses, the instantiation principle is fatal, for instance, for trying to compare the consequences of alternative viewpoints with alternative ontologies. I am talking above about reference ontology type uses.

[Dr. Adam M. Goldstein PhD MSLIS]

On Oct 26, 2011, at 11:17 AM, Stefan Schulz wrote:

>>
>> If an apple can have a "redness" at a time point, why can a process not
>> have a "duration"?
>
> I think there is a consensus that all processes have a duration.
> Whether we name the duration of a process a "process quality", and the
> relation between the duration and the process "inherence", is just a
> labelling issue and therefore a matter of convention. There may be
> good reasons to
> be more restrictive with the scope of relations in order to avoid
> disjunctions in domain and range restrictions. I can therefore live
> with a convention that restricts the range of "inheres_in" to
> independent continuants.
>
> What is still open for me:
>
> 1.) How to relate processes with their durations, e.g. by a relation
> "has duration" ?
>
> 2.) Whether the alleged process qualities can be expressed by
> combining the duration of a process with the qualities of its
> participants.
>
> 3.) Given 2.) is true, whether process qualities such as in phenotype
> ontologies can be allowed as a kind of "second class entities", for
> the sake of simplicity. I am sure that if BFO disallows process
> qualities, BFO users won't care and introduce whatsower relation in
> order to, e.g. , connect " course of disease" with "chronic".

Briefly: I do agree with Pierre G when he points out that we cannot declare that processes have properties in the sense that entities (continuants) do, because processes do not have dependent continuants. This is integral to the structure of the BFO and is warranted by the metaphysical view the BFO embodies. I do think that if we are going to take the BFO seriously, we can't permit any kind of inherence of properties in processes.

Similarly, the idea that we categorize heartbeats, propellers as their blades go around, metronomes, and the lot as a kind of rate, isn't right. Think about it: what commonality, besides having some activity measured in beats per minute, do all these things have in common? This is a miscellaneous group of things if there ever were one! This is, I suppose, an onbtological intuition that others might not share. In any case, think about if you had a big box of things that included a bunch of things that could be measured in beats per unit time in some way. You would group the musical time-keepers together, rather than saying "hey, this is already a coherent group of things."

(I think Albert Goldfain made ths suggestion earlier in this thread, but I can't find his email. I mean not disrespect by flatly disagreeing! I am writing in the spirit of spirited exchange of ideas.)

I made a suggestion earlier in this thread, that goes as follows.

There is a discintion between determinates and determinables. The determinables are things that can be further specified by some kind of additional characterstic, drawn from a set of characteristics. For instance, color is a determinable; red is a determinate. A red ball has a color, which is determinable, by one of the colors.

Reading David Armstrong recently for my metaphysics class, I came across the following idea. The determinates, not the determinables, are the real entities, the determinates being the real things. So, any determinable must be further specified, in order to describe the reality of the thing.

In this case, this works as follows. First, forget about the heart, because that is a continuant. Maybe it participates in a heart beating process or something. I don't know. But we do have heart-beating process. Maybe it's more intuitive to say: we have beating hearts. But, since every thing that's beating must beat at a particular rate, "beating heart" is not the kind of ontological fundamental process we want; for that, we have beating heart, 60 bpm, and so on, for all of the different rates per minute. If we hold the instantiation principle, as we do on the BFO, then we don't say that there is a category for every rate a heart could beat; we only have such categories for the rates at which hearts actually have beaten.

I think that a dynamical description might be described similarly. There are hearts-beating-declining, 2 bpm/m, and similarly for the rest of the rates of increase or decrease.

I think this really does capture the intuition that we want to put hearts beating a different rates or decreasing or increasing at certain rates of change into different categories, and that what hearts beating is most essentially a kind of process, something that's extended over time, perhaps starting at the start of an organism's life, ending with its death.

Regarding durations, I don't know how this would work out. 10-year-long-heart-beating is a different kind of heart-beating in contrast with 25-year-heart-beating. Maybe not so bad? Allowing for all other factors, the mean heat-beating-duration among smokers is lower than that of non-smokers.

I'd be interested to hear Pierre and Barry's suggestions about this, and about the general principle of determinates being the ontological foundation for determinables.

There as a really great paper about this, I think about the different sense of "is_a", but I cannot for the life of me find it, and if anyone knows about such a paper, send me the information.

As an aside, I think that the BFO top-down approach should not be the only one, and that probably its utility has been greatly overemphasized. Instead I think that an approach I call ontological naturalism is better. I am currently finishing off a manuscript I intend to submit to Applied Ontology about this.
>

[Jessica Turner]

Hi Barry, what is this meeting Nov 10 and 11, and will those of us following this discussion but not at that meeting get a chance to see the presentation/paper?  It seems like that's the nub of the matter--what errors would be introduced if we classed both "red" and "rate" as qualities, or red was quality1 (limited to continuants) and rate was of type quality2 (limited to processes), is where I'm still getting hung up. (I don't fear schism from BFO, but I want to know what the metaphysical implications are, and what logical errors I'm going to be facing if I do drift from the one true way... ;)

Thanks,

Jessica

On Tue, Oct 25, 2011 at 5:50 AM, Barry Smith <phis...@buffalo.edu> wrote:

The argument, all along, is that when we make predications concerning processes we are not referring to some extra entity, as we are when we assert, for example, that a continuant has a suntan or has a disease; rather we are asserting that the process is of a certain type, for instance it is of the type beating process, or of the type beating process with rate 60bpm, and so forth.

We have finally (I hope) found a way to document this idea coherently, and I hope that it will be in good enough shape to present at the meeting on November 10-11.

BS

On Mon, Oct 24, 2011 at 4:49 PM, Chris Mungall <cjmu...@lbl.gov> wrote:

Can someone remind me why qualities of processes (e.g. rates) are disallowed? Is there documentation e.g. on a wiki somewhere?

Let's just say that BFO were to allow qualities to inhere in processes. There would be many pragmatic advantages. Modeling would be easier and more in line with scientists' intuitions.

What bad things would happen? Would interoperability in some way be hindered? Am I missing something?

--
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.

--
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.

[Pierre Grenon]

Hi Melissa,

thanks for your reply! Just a brief, superficial follow up for now.

There are two main difficulties with case 1. The first one, as you
rightly point out, is the pervasive type-instance issue in the level
of discourse. Another is the epistemic dimension and contextual
character of some of the assertions made. For the later there are
knowledge representation approaches which BFO could live well with but
that are cumbersome in the world of RDF and OWL (and that is true of
anything, not BFO in particular). There should be a trade off allowing
to do something simple in a way that will remain consistent when it is
perhaps done right... The simplest thing to do is to abstract the
epistemic dimension, unfortunately this is very wrong and also it may
not prove that useful, although still somewhat useful...

With provisions made for the type/instance issue, the other cases are
comparatively easier. For example, 4, the asynchronicty of processes,
seems to directly translate into BFO as a case of spatio-temporally
scattered processual --- it's not entirely straightforward though and,
of course, the current OWL encoding doesn't provide the notion
explicitly.

Comparing onsets is straightforward when dealing with 2 processes
(case 3 i still seem to understand). Just a definitional extension
comparing the occurrences of these processes should be enough, it's
built into BFO but it's not in the OWL encoding (which is true of 95%
of BFO anyway). For this, as is standard, BFO can live with some sort
of Allen calculus... The need for an OWL encoding of certain such
definitional extensions has been raise. So this is more water for the
mill.

I'm still a bit puzzled by 3. But let me come back to 1 and 3 when
time permits.

I know this is cryptic but I have to take a rain chek for now...
perhaps we could start a new thread for these concrete questions you
raised.

cheers
p

[Pierre Grenon]

On Wed, Oct 26, 2011 at 5:07 PM, Jessica Turner <jtu...@mrn.org> wrote:
> Wait, if as Pierre suggests below, there could be a relationship between
> processes that captures the idea that one process has a faster rate than
> another process, and yet processes can't have qualities like rates, how
> would that work?

such a relation would be primitive in the relevant context. One would
just assert the relationship and, all things being equal, querying
would be simple retrieval.

> What should you be able to infer from knowing that that
> relationship holds between the two processes, if it isn't the naive idea
> that the processes each have a rate and one of them is greater or faster
> than the other?

well, yes, you could decide to interpret the relationship as meaning
this and define it so. For example, as far as I understand, that'd be
the way to bridge between PATO as it stands and a rate-less BFO
compliant version of it.

On the other hand, the stake of having rates in the first place, in
the present use case, that is, seemed to be in being able to infer the
relationship between processes. Also, the relationship between
processes seemed to be that which was of interest for the purpose of
annotation.

So, even by the standard of the intended use case, it is simpler to
have just the relationship. This is the case whether or not rates
exist. Compare:

A. Annotation stage: assert a relationship between processes

B. Annotation stage:
1. instantiate rate (first process)
2. instantiate rate (second process)
3. assert a relationship between rates

Application stage: try to infer the relationship between processes

A is simpler and in fact depending on circumstances may also be the
only feasible option.

There's no sweeping argument one way or another here, I'm just
pointing out that a tentative rate-less approach is both simple and
sufficient to answer a pragmatic use in the present case.

cheers
p

[Pierre Grenon]

sorry for the afterthought, i meant to ask. Melissa, what do you do at
the moment for the cases you raise?

[Phillip Lord]

> It's similar to another common sensical move suggested long ago, i.e.
> have an overarching category of regions in BFO, because, well, the
> classes of spatial regions and of spatiotemporal regions are both
> named after some type of regions, ain't they? In the present
> methodlogical context, this is not a compelling argument, nor more
> than to say that there is an overarching category of ducks, ducks and
> rubber ducks.

While the analogy with Rubber Ducks is entertaining, it is rather broken. The
reason for suggesting an over-arching concept of region, tying together
Spatial and Temporal Regions in BFO is not an example of confusing lexical
structure with ontological. It stems instead from the fact that this form of
model has been used in physics for the last hundred years. In fact, the
adoption of this model produced one of the more profound and vivid shifts in
our understanding of the nature of physical reality; it has lead to many
predictions which are both testable and have been tested. It has stood the
test of time.

We can set this against BFO. Your conceptualisation leads us to the conclusion
that a SpatialRegion cannot have a length. We can, I think, safely set aside
whatever clever philosophical wordplay you have used to convince yourself
otherwise, because we have a strong counter-example. Maybe, if you had
listened to those who suggested an over-arching concept of region, you would
not have made this mistake.

Now, the fact that BFOs model is very poor fit to physical reality in this
area, is not a huge problem. BFOs conceptualisation is only really meant to
support biomedical use cases; for this purpose, no one uses spatial regions
anyway. However, it is the same problem, the idea that dependent continuants
can only inhere in independent continuants that causes problems with your
concept of a process. And this is a serious problem, because people are trying
to use the concept of a process.

If we consider, for example, the amount of a solute, the concentration of a
solution and the rate of change of that concentration as a chemical reaction
occurs. You may consider that, for instance, the concentration and its rate of
change are qualities of the solution. But, this is not enough. We have to have
a relationship between the rate of change and the concentration. They are
clearly directly related, as is the rate of change related to the process
causing that change.

Perhaps it is time to take a step back, and realise that it is possible that
science tells us more about physical reality than philosophy. You have made a
mistake here, and BFOs own rate of change in addressing this mistake has been
glacial.

[Cristian Cocos]

On 10/26/2011 5:39 PM, Jessica Turner wrote:

It seems like that's the nub of the matter--what errors would be introduced if we classed both "red" and "rate" as qualities, or red was quality1 (limited to continuants) and rate was of type quality2 (limited to processes), is where I'm still getting hung up. (I don't fear schism from BFO, but I want to know what the metaphysical implications are, and what logical errors I'm going to be facing if I do drift from the one true way... ;)

My guess as to what the answer to this one might be is that you would not be committing any logical errors, but ontological errors, errors of adequacy: In the ultimate analysis, the ontology would not accurately/adequately reflect reality. Anyway, I'm curious about this one myself, as, to my mind, everything that is currently captured via "quality" can very well be captured via "types" (however fantologist this might sound :-) ). Thus, interestingly, a fantologist answer seems to be adequate for processes/ocurrents, but not adequate for continuants: for continuants you must have qualities, whereas for occurrents everything has to be done fantology-style.

C

[Adam M. Goldstein]

On Oct 28, 2011, at 11:01, philli...@newcastle.ac.uk (Phillip Lord) wrote:
>
> Now, the fact that BFOs model is very poor fit to physical reality in this
> area, is not a huge problem. BFOs conceptualisation is only really meant to
> support biomedical use cases; for this purpose, no one uses spatial regions
> anyway.

The BFO, rather than being an upper-level ontology, is a domain ontology that is an extension of the Gene Ontology.

------------------
Adam M. Goldstein PhD, MSLIS
--
z_calif...@shiftingbalance.org
http://www.shiftingbalance.org
http://www.twitter.com/shiftingbalance
--
http://www.itis.gov/servlet/SingleRpt/SingleRpt?search_topic=TSN&search_value=180621
--
(914) 637-2717 (msg)
--
Dept of Philosophy
Iona College
715 North Avenue
New Rochelle NY 10801
http://www.iona.edu/faculty/agoldstein

[Melissa Haendel]

Hi Pierre,

Not entirely sure what you mean by "what do we do" but if the question is what ontologies/relations do we use for phenotype annotations, we basically compose them using a compendium of ontologies: continuent entities largely from anatomy ontologies or GO cellular component, qualities largely from PATO, spatial modifiers from spatial.obo, processes largely from GO biological process. Note that when I say "qualities" I don't mean them in a BFO sense, but rather the content of PATO (which is a mixed bag and not BFO compliant) and also enabling comparative relational statements (eg. bone A fused_to bone B in species 1 relative to species 2). 

Cheers,

Melissa

[Nico Adams]

One thing that hasn't been mentioned to my surprise is that
ontological frameworks - and this is no defense specifically of the
BFO or any other upper ontology for that matter - can be a useful
guide to clarifying the ontological status of entities - and to date I
see that as their best contribution. Because after all, that is one of
the main reasons why we construct ontologies: to clearly say in
explicit and formal semantics what it is that we mean when we refer to
an enity, talk about it etc...so that both machines as well as humans
can make decisions about that object. Clear semantics is of importance
here. Ontological frameworks help, but also impose limitations due to
ontological commitments. Which is why scientists, when they do commit
to using one of these frameworks, actually have to understand it's
inner workings....so I can't completely follow Pierre here either.

I also do not think that the interoperability argument has been shown
to hold and seems more like an assertion.

As to doing science by consensus: I note that for a long time Galen's
view of anatomy was accepted scientific consensus. It took Vesalius to
break away, fracture and move medicine forward. Similar thing in
astronomy: geocentrism was the accepted community view until that
annoying Copernicus came along. Science has never really done too well
when it tries to develop on community consensus.

All the best,

Nico

On Thu, Oct 27, 2011 at 12:27 AM, Robert Hoehndorf <leec...@leechuck.de> wrote:
>>>>>> "CM" == Chris Mungall <cjmu...@lbl.gov> writes:
>

>    CM> For one thing it would create a schism rendering it harder to
>    CM> interoperate with ontologies that do use BFO, which most of us I
>    CM> think would count as a Bad Thing.
>
> Can you point me to one case where the application of BFO has lead to
> "interoperability" between ontologies that would not have been possible
> without it? I am quite interested in what kind of "operations" were
> enabled through BFO's use.
>

>    CM> I believe in moving forward based on consensus, rather than
>    CM> splintering and forming factions, based on one person's
>    CM> needs. I'd hope that if there is a consensus amongst users that
>    CM> some part of BFO needs fixing then it would be fixed - this is
>    CM> no different from any other ontology.

>
> This seems to be the core of the debate: will biomedical resources,
> annotation guidelines, ontologies, etc. be made to change so that they
> accommodate the view on the world imposed by one philosophical theory
> called BFO, or should there be an evidence-driven, empirical way to
> connect the existing resources and make the relevant distinctions
> explicit? Sometimes, it seems to me that BFO and parts of OBO seem to be
> taken the first route: make a philosophically-motivated decision, and
> convince people to change whatever they are doing to accommodate it. If
> this will be the most successful strategy is still an open question, I
> think.
>

> Rob.

[Pierre Grenon]

On Wed, Oct 26, 2011 at 2:54 PM, Nico Adams <nic...@gmail.com> wrote:
> One thing that hasn't been mentioned to my surprise is that
> ontological frameworks - and this is no defense specifically of the
> BFO or any other upper ontology for that matter - can be a useful
> guide to clarifying the ontological status of entities - and to date I
> see that as their best contribution. Because after all, that is one of
> the main reasons why we construct ontologies: to clearly say in
> explicit and formal semantics what it is that we mean when we refer to
> an enity, talk about it etc...so that both machines as well as humans
> can make decisions about that object. Clear semantics is of importance
> here. Ontological frameworks help, but also impose limitations due to
> ontological commitments. Which is why scientists, when they do commit
> to using one of these frameworks, actually have to understand it's
> inner workings....so I can't completely follow Pierre here either.

I don't entirely follow why we are not on the same page. My two cents
here is that relevance and cognitive salience are contextual and that
I'm not convinced that to a certain degree it is more than a
distraction for scientists to wonder about the inner working of, say,
a formal notion of substance when they are dealing with very concrete
objects and phenomena. This may be oversimplistic and clearly it ties
to who is doing what and what they are trying to do... but let's try
to avoid casuistry. Perhaps also the word 'scientist' is used to
include bioinformaticians, bringing into the picture people who don't
all worry about, or not only about science (yeah, yeah don't throw
stones at me, I'll admit there's a grey area.) That's only for the
human part...

> I also do not think that the interoperability argument has been shown
> to hold and seems more like an assertion.

Agreed. It's still more of an hypothesis that is part and parcel of
the motivation and the methodological disposition surrounding work on
certain ontologies, BFO in this case and, in this case too, in the
context of OBO. The assumption is that it is worth shooting for a
unifying formal ontology. That's a gamble. Whether it is useful at
all, how useful it is and how it compares to alternative approaches is
a broad question that nowadays involves more than technical aspects.
For the technical side, the proof of concept has been made for a long
time with Cyc (there's a system with a somewhat coherent core formal
ontological approach expanded in various directions to handle
different domains---that's one reading, caveat exist). Some of the non
technical benefits, again, concern division of labour. Yet it's clear
the collegial dimension, in particular, is not precisely without
problems of its own.

> As to doing science by consensus: I note that for a long time Galen's
> view of anatomy was accepted scientific consensus. It took Vesalius to
> break away, fracture and move medicine forward. Similar thing in
> astronomy: geocentrism was the accepted community view until that
> annoying Copernicus came along. Science has never really done too well
> when it tries to develop on community consensus.

Whichever the model of science, BFO does not aim at operating at the
level of meddling and adjudication. While I'm happy for this claim to
be stigmatised as naive, it is hard to read out the decrying of BFO's
nefarious influence on virtuous science as more than discontent with a
certain methodology, not least its realist tincture.

cheers
p