Re: blog: semantic dissonance in uniprot

On Mar 27, 2009, at 4:23 AM, John Madden wrote:

>>
>>> So, here's how I'd do this. Introduce a property linking a protein  
>>> to  something (which might be anything from a piece of text to a  
>>> protein) called sameProteinAs. Its reflexive and transitive but  
>>> might not be symmetric (though it probably is when the value is  
>>> itself a protein). It is NOT substitutive. It means, roughly, that  
>>> its value either is, or has as its main topic, the same protein as  
>>> the argument. It is a mixture of sameAs restricted to proteins and  
>>> seeAlso restricted to cases where the topic is a single protein.
>>
>> The "something" to which you link could even just be a blank node,  
>> Basically, if I understand you correctly, it's just a hypothetical  
>> tertium quid, that you might later abandon or declare to be devoid  
>> of any useful meaning. Or perhaps better, it's a collection that  
>> collects things that somebody thought were "similar" to each other.  
>> So if it's a class or set, it's a set whose intension is defined by  
>> some human opinion, not a class that makes any claim on being like  
>> a natural kind.
>
> Uh-oh. Rereading your proposal, I see I misinterpreted it. Apologize.
>
> Would this be like a property defined by a rule? e.g.
>
> If (subject type protein) and (object type protein)
> 	then sameAs
> else seeAlso.
>
> or maybe
>
> if (type(subject) equivalentClass type(object))
> 	then sameAs
> else seeAlso.
>
>


Yes, though those rules are overly strong: I wouldn't want an  
automatic system to be able to infer sameAs unless someone had  
explicltly asserted it. But this gets the general idea, yes.

I think its fine for the SWeb to include 'weak' semantic links that  
don't (yet?) have tight definitions that can support machine  
inference, but still convey useful information to users and maybe even  
tool developers. (I know saying this runs the risk of opening the old  
'social meaning' can of worms, but those worms aren't going to go  
away :-)

Pat

------------------------------------------------------------
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