Re: blog: semantic dissonance in uniprot

On Mar 26, 2009, at 6:53 AM, Phillip Lord wrote:

> Oliver Ruebenacker <curoli@gmail.com> writes:
>>  Besides, how do we know it's wrong? Two species can have the same
>> protein for different functions, right?
>>
>
> Depends how you define "same". This is the crux of the problem.
>

No, really, its not. Defining 'same' is trivial. A is the same as B  
when there is one thing, called by both names 'A' and 'B'. Now I  
realize this sounds circular (and trivial), but the key point is that  
if you are using a logic-based language (such as OWL or RDF) _at all_,  
then you must have some notion of what the things are that you are  
talking about. You are using names (URI references , IRIs), and names  
denote things. If you are using OWL, you are thinking about classes of  
things. You must have some things in mind when you do this. Whatever  
those things are, being the same means being the same one of those.

OK, I know that 'identity criteria' are often problematic (philosopher  
talk) and that its easy to fall into use/mention confusions (such as  
between database records and the things they describe) and generally,  
getting the ontology right isn't always easy. But when this happens,  
don't blame owl:sameAs. If you don't know what your ontology is  
talking about, or you cannot even say in English what it is that you  
want to say in Formaleze, your problems are much more fundamental and  
much pressing than anything to do with sameAs, and they won't be  
solved by making sameAs fuzzier or less rigorous, or whatever. This  
is  a symptom of a deeper malaise.

These formal languages require one to have a sharp and clear sense of  
what it is that one is talking about, and to stick to it. For many  
purposes (eg see OBO) this what they are for, to impose a kind of  
global coherence on a body of discourse. Making them fuzzy and  
contextual and un-"rigorous" would be like making banks use vague  
arithmetic in order to solve the financial crisis. Sorry of this means  
that your thinking is too vague to be OWL-ized just yet. Come back  
when you get your thinking sorted out, and maybe we'll be in a  
position to help you.

Pat


> Phil
>
>
>

------------------------------------------------------------
IHMC                                     (850)434 8903 or (650)494 3973
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes

Received on Thursday, 26 March 2009 14:49:33 UTC