Re: homonym URIs (Re: What if an URI also is a URL)

>Pierre-Antoine Champin wrote:
>>I think (as Pat, if I read him correctly) that punning/overloading can
>>not be avoided. I would add that it can be deliberate, for practical
>>reasons (e.g. e-mail adress / person, predicate / function), but it can
>>also be *unintentional*. Let me explain :
>I think this sort of punning is a red-herring, because it isn't really a pun.
>If I put an email address into a context where the expected input is 
>"email address of person in question" then there is no pun; it 
>identifies a mailbox. That the mailbox is in turn being used to 
>identify a person doesn't make it a pun.

Its a pun in the relevant sense if its the same URI being used to 
refer to/identify more than one thing. True, you can always insist 
that one of them is 'primary' and the others derived from it (this is 
how the Common Logic semantics works, in fact, where the individual 
is primary), but that only makes sense if we all agree ahead of time 
which is the primary meaning.

>>we keep using the same word for slightly different things (e.g. a city
>>as an administrative entity or as a populated location), as long as the
>>difference between them is not relevant to us. The same will be true of
>>URIs that we will create and put in RDF. We can not expect everybody on
>>the web to require the same level of detail on every part of the world
>>about which they make RDF statements.
>If you have a URI that means "London; the administrative entity" 
>then it means just that.
>If you have a URI that means "London; the populated location" then 
>it means just that.
>If you have a URI that means "London; a populated location which was 
>made an administrative entity" then it means just that.

But this assumes that we have made all the relevant distinctions 
ahead of time. Indeed, then it is easy, as long as we all agree. But 
this will never happen, since we never can have made all the relevant 
distinctions. Someone may just have a URI for "London", and not have 
even thought about the (administrative entity)/(populated location) 
distinction. Even if it is pointed out to them, they may refuse to 
make it, saying that they don't care about such hair-splitting; and I 
think they should be able to do that. There are distinctions, which 
are debated with passion among ontologists, which are close to 
incomprehensible to anyone but a professional philosophical logician, 
yet to get them wrong can wreck some ontological reasoners. Even such 
apparently secure things as GPS references have multiple layers of 
ambiguity. In some military settings, the very same assertion made by 
sources at different command levels may have different military 

>You can go from one to the other if you know the relationship 
>between them. If I start thinking of London - populated location and 
>the context makes it apparant I should be thinking about the 
>administrative entity then the mental switching of gears is doing 
>exactly that.
>There are clearly problems if a system doesn't know about that 
>connection between the three different concepts above, but if it's 
>meant to know about them then the problem isn't in the URIs.

The problem is that no system can be expected to know about *all* 
distinctions that *all* other systems may need to make. So almost any 
act of reference is going to seem ambiguous to someone.

>>An intuition is that owl:sameAs may be too strong a statement
>>in a context where URI can be ambiguous.
>I think that's definitely true.

Yes, logical equality used on the Web means "equal everywhere, to 
everyone", which is a very strong assertion.


IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell

Received on Wednesday, 13 June 2007 17:12:11 UTC