Re: blog: semantic dissonance in uniprot

On 26 Mar 2009, at 22:02, John Madden wrote:

> Hi Peter,
>
>> It would be infinitely better than seeAlso where you have no idea  
>> what
>> the intent is, other than it isn't owl:sameAs or anything else in
>> between.
>
> Of course, you could create a property ex:similarTo in your own  
> namespace, and then declare that
>
> 	ex:similarTo rdfs:subPropertyOf rdf:seeAlso .
>
> Then you could just make some English language usage recommendations  
> about ex:similarTo.

Of course, there are always at least three options:

	1) Overload existing term
	2) Introduce a user defined new term
	3) Introduce a new built-in term

Since we're in the land of 2 and 3, some key questions are:
	a) How do we define the term? (Natural language, rdf or owl axioms,  
operationally)
	b) What is the definition (in particular, how tightly do we define  
it; what part of the infrastructure should be sensitive to which parts  
of the definition)

I think signaling intent, at least the intent to say something more  
specific than seeAlso, is good. But I also think we can do better on  
the definition.

Cheers,
Bijan.

Received on Friday, 27 March 2009 08:31:56 UTC