At 12:44 PM 3/23/2008, Pat Hayes wrote:
>At 4:45 AM -0400 3/20/08, Barry Smith wrote:
>>At 12:57 PM 3/20/2008, Pat Hayes wrote:
>>>At 9:03 AM -0400 3/20/08, <phismith@xxxxxxxxxxx> wrote:
>>>>On Thu Mar 20 2:34 , "John F. Sowa" sent:
>>>>
>>>>>For any product, including an ontology, the best recommendations are
>>>>>the reviews and reports from users that are recorded in the metadata.
>>>>>As the reviews accumulate, prospective users can decide for themselves
>>>>>which ontologies are best suited for their purpose(s).
>>>>
>>>>I think democratic ranking (the wisdom of crowds) is indeed valuable for
>>>>refrigerators and other similar products. Hence the success of
>>>>ranking systems on
>>>>eBay and amazon.com. But it is surely of less importance in
>>>>scientific contexts
>>>>-- we would not decide on which interpretation of the equations of quantum
>>>>physics to accept by taking a vote of users. Since the OBO
>>>>Foundry ontologies are
>>>>built by scientists, to support scientific research, it is not
>>>>clear that they
>>>>are to be treated as products.
>>>
>>>This is where I part company with Barry, and indeed where I
>>>believe that the very idea of controlling the contents of an OOR
>>>(noting that the first O means 'open') needs to be examined very,
>>>very carefully. Of course we would not argue that majority voting
>>>should be used to choose scientific theories; but ontologies, even
>>>those used by scientists, are not themselves scientific theories.
>>>The OBO Foundry is quite clear, in its own documentation, that the
>>>basic ontological assumptions on which it is based are ultimately
>>>philosophical decisions, not scientific ones.
>>
>>The actual work of the Foundry in maintaining its ontologies is 50%
>>of the time focused on getting the science right. I think this
>>percentage will rise, as the ontologies themselves become more mature.
>
>Fair enough. I do however see a lot of email traffic devoted to
>questions of which things have to be put into which philosophical
>categories. I don't accept that calling something a continuant is
>doing science. (01)
Of course. (02)
>>> Such assumptions most emphatically do not have the force of a
>>> scientific theory, even when the ontologies constructed according
>>> to them are being used by scientists.
>>
>>So anything goes, eh?
>
>Not necessarily. (It is revealing, I think, that you seem to think
>that anything other than kosher science is anarchy.) (03)
If we are building ontologies to support kosher science, then the
ontologies should be compatible with the results of kosher science
(and evolve as we learn more from doing kosher science and from
discovering how ontologies can best help with kosher science).
The point of ontologies, in all of this, is to help solve an
increasingly severe problem of data siloing (I am tempted to call it
data silage), by providing convergence on single, algorithmically
useful representations of the domains from which data are collected.
For this to work ontologies have to be as far as possible unique.
(Part of this is a bit like driving on the same side of the road, and
some of the most successful bits of the OBO Foundry methodology are
non-controversial rulings of this sort -- e.g. do not mix singular
and plural nouns but use singular throughout. (These bits do not
cause pushback in the form of email traffic.) (04)
Hence where you characterise our differences as: (05)
>I think the main differences between our attitudes is that you see
>ontologies as representing a distillation of a common insight,
>ideally amounting to something close to a universally agreed
>scientific theory; whereas I see ontologies as much more like useful
>pieces of software. There are many text editors, and they all have
>their strengths and weaknesses, adherents and detractors. No size will fit all. (06)
I would say: (07)
I see scientific ontologies as being involved in a necessary
convergence on a single scientific-evidence-based representation for
each domain. From these, reference ontologies, many application
ontologies (your 'products') are being formed; but the reference
ontologies are needed to keep the products in line with each other
(otherwise the silage problem arises once again). (08)
>>>And any such implication of 'scientific' authority must be
>>>examined especially carefully when the, er, foundry is controlled
>>>by the philosophers themselves (09)
>>One philosopher, one computer scientist, two computer
>>scientist-biologists, one immunologist, one world-class geneticist.
>>
>>>, and its gatekeepers are mandated to only allow ontologies which
>>>conform to the somewhat arbitrary philosophical views of its
>>>founders (for example, by requiring consistency with a single 'base'
>ontology).
>>
>>Actually not -- the single base ontology we started with has
>>already been modified because it did not fit the science.
>
>I applaud that kind of a move, but it does not really counter my
>point. Would you allow an ontology which explicitly denied the
>continuant/occurrent distinction? (010)
It is an empirical question. When we see a good one for a specific
area of biology we will examine it, see whether it works better than
what we have. (011)
>>> I do not mean this to be a criticism of OBO itself, but I do
>>> claim that OBO hardly qualifies as anything like an "open"
>>> ontology repository. In the contrary, in fact: it is quite firmly
>>> closed to an entire approach to ontology construction which,
>>> while successfully deployed elsewhere, happens to not conform to
>>> the philosophical views that Barry has so nobly defended in so
>>> many publications. (012)
OBO is open in the sense you require, I think. The OBO Foundry is
constrained, to some degree, in the sense you describe. But I am not
by any means all powerful in the OBO Foundry. Many views were
accepted by what turned into the main OBO Foundry ontologies before I
came along. Some views I (and others) have to argue for strenuously.
Sometimes I (we) lose those arguments. Evidence-based ontology is
involves getting many things right which are not addressed at the
level of OBO/OWL/RDF formatting questions.
BS (013)
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2008/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2008
Community Portal: http://ontolog.cim3.net/ (014)
|