RE: Disambiguation; keeping the "U" in "URI"

> I am quickly coming to the conclusion that there are only two
> options
> left:
> 
>   1. Don't use http:// URIs in RDF you author/process/trust

That's not really an option.  Next you will have people using mailto: URIs to refer to ten different things, and all claiming they have a right to do so, and you will be in the same situation.  Any URI scheme you can come up with can be attacked by people who say "but what if I want to say 'bad' when I mean 'good'?"

>   2. Don't trust URIs to refer to distinct entities

All this is suggesting is that we stop using URIs to identify things.  So I guess we should come up with a different scheme for coming up with identifiers for things.  Maybe we will call it Universal Thing Identifiers, or UTI.

> It seems that giving up 2 defeats the purpose of RDF.  So, are we

I think it is not so much RDF that suffers, since it also eliminates usefulness of N3, bare naked triples, or any other semantic information.  Essentially #2 defeats the purpose of *URIs*.

If you want to make assertions about things, you need a uniform way to identify them.  If we didn't have URIs, we would have to invent them.  I really don't think it is such a problem to just use URIs as identifiers, like they were intended (rather than as hints).  The only people who will have trouble merging and integrating their data into the semantic web will be the ones who think that:

http://www.mysite.org is sometimes equivalent to:
(http://www.mysite.org dc:Creator)

No, we can't stop people from asserting that a web page is a car, but that doesn't mean that we have to jump through hoops to accommodate those people or worry about having to interop with them later.  It's not as if there is any working interop (at the metadata level) or semantic web legacy that we have to avoid breaking.  The train hasn't even left the station.

Received on Thursday, 25 April 2002 01:21:21 UTC