From: Michel_Dumontier <Mic...@ca...> - 2010-08-29 03:55:27
|
> -----Original Message----- > From: Barry Smith [mailto:phi...@bu...] > Sent: Saturday, August 28, 2010 7:57 PM > To: obo...@li...; obo- > di...@li... > Subject: Re: [Obo-discuss] OBO Foundry ontologies and conformance to > OBO Principles > > At 07:44 PM 8/28/2010, Adam M. Goldstein wrote: > ><mailto:phi...@bu...>On Aug 28, 2010, at 14:57, Barry Smith > ><<mailto:phi...@bu...>phi...@bu...> wrote: > > > >> > >> > >>>Michel_Dumontier wrote: > >>>As members of the ontology community (in terms of design, > >>>application and evaluation), we are actively engaging so as to > >>>ensure that the ontologies produced are sensible and maximally > >>>useful. We really need better justification for the realist criteria > >>>in scientific ontologies (other than "we should stick to entities > >>>that exist"). Our discussion here and additional investigation into > >>>the contents of databases and published articles should convince you > >>>that this is not sufficient. > >>> > >>>We are unconvinced that the solution lies in building ambiguously > >>>defined "application" ontologies as this is both unclear and unsound > >>>(consider: if every molecule must have been demonstrated to have > >>>instances, and C16 is kind of molecule, then C16 must have been > >>>demonstrated to have instances). > > > >It is pointed out below that the OBOF is not committed to what > >Michel says above, about C16. I wonder, is the OBOF committed to the > >following, which I advance as a first stab at a precise formulation > >of the instantiation principle? > > > >(1) A term T, for a universal U, in an ontology O, satisfies the > >instantiation principle for terms if and only if U has at least one > instance. > > > >(2) An ontology O satisfies the instantiation principle for formal > >ontologies if and only if all of its terms T satisfy (1). > > There is an important reference to evidence, here missing. > Evidence is, be it noted, not always verdical. > > > >I don't know if "in reality," should be appended to (1) in order to > >represent the OBOF's intentions. I don't know how else "instance" > >ought to be understood, so it's probably superfluous. But it does > >seem as though "instances in reality" is in dispute. > > > >What an instance is needs much more clarification > > a is an instance of ; but at any rate, I don't think what's at issue > here is what an instance is, but whether (2) is a necessary condition > for ontologies---that is, whether instances are required for all > terms in an ontology. > > Your requirement would imply that the authors of ontologies are > infallible beings. Weird! Adam is clearly restating the obvious question that we keep asking over and over again! are instances required for every term introduced in an ontology? If so, then we can actually formulate a meaningful requirement for OBO Foundry ontologies -> that every term in an OBO Foundry ontology must be documented with a published reference to the experimental evidence of instantiation (as you state below). > > > >>Part of the OBO Foundry effort is to document carefully the > >>principles that we recommend. Providing such documentation is a > >>difficult business, and we have clearly, in light of what Michel > >>writes here, to document clearly what the Principle of Instantation > means. > >> > >>In brief, it says that a term should be included in a reference > >>ontology only if there is experimental evidence that instances to > >>which that term refers exist in reality. > > > >What I personally would like to know is, what is intended by "exists > >in reality" here? In much of the discussion on this list, I get the > >sense that it's intended in the context of the OBOF to mean > >something like "there is some mind-independent entity." The view is > >that ontologies ought only to have terms for universals if those > >universals have instances that are mind-independent entities. > > Apart from domains such as psychology, where the entities in question > are allowed to be mind-dependent, yes. > > >I think that what some people don't understand, or don't agree with, > >is this commitment to mind-independent entities. > > Did the planet exist 1 millions years ago? > > >What Michel is asking for, I think, is some argument for that > >commitment, for some justification for adopting it. > > Because it has been found useful, above all by the Gene Ontology. And the UMLS, NCI-T and many other terminology resources and ontologies have their own methodologies that are deemed useful, so that's not an interesting argument. > >There are some well-known, well-motivated, and highly plausible > >alternatives to this form of realism in the literature dating back > >at least as far as Kepler. > > Yes. Defenders of these plausible alternatives are welcome to build > ontologies to fit. > BS Again, weirdly, the position we are advocating entirely contains and goes beyond yours. You need to better justify your position as to why we should exclude a broader approach for the design of ontologies that provide meaning to scientific terms. m. > > > > These include varieties of pragmatism and anti-realism of the kind > > advocated by Duhem and van Fraassen. The choice of this kind of > > realism over these alternatives has, so far as I can tell, never > > been explicitly defended in the formal ontology literature or in > > the online discussions. > > <http://monist.buffalo.edu/ToC/77_1.html>http://monist.buffalo.edu/ToC/ > 77_1.html > > BS > > > > ----------------------------------------------------------------------- > ------- > Sell apps to millions through the Intel(R) Atom(Tm) Developer Program > Be part of this innovative community and reach millions of netbook > users > worldwide. Take advantage of special opportunities to increase revenue > and > speed time-to-market. Join now, and jumpstart your future. > http://p.sf.net/sfu/intel-atom-d2d > _______________________________________________ > Obo-discuss mailing list > Obo...@li... > https://lists.sourceforge.net/lists/listinfo/obo-discuss |