Re: Explicit Disambiguation Via RDF bNodes, more Process

Joshua Allen wrote:

> Is it fair to say that your proposal is:
> 
> 1) Assume that a URI identifies a unique thing (and an http: URI identifies a "web page")
> 2) Assume that it is OK to merge assertions that use the same URI
> 3) Acknowledge that sometimes you might get metadata where a single URI is used inconsistently.
> 4) Metadata with inconsistencies may or may not be able to be merged, depending on information available and techniques used to merge it.  It may be possible to merge it, or it may simply be necessary to ignore it.


Yes - I like this summation. If merged metadata contains inconsistencies 
we basically have 4 choices (as I see it):
1) ignore the bits that are inconsistent
2) do some sort of inferencing/ probablistic heuristics to determine a 
"prefered" (or more plausible) interpretation (and ignore the 
interpretation we don't like)
3) in the case of ambiguous identifiers - create 'virtual' nodes that 
disambiguate 2 objects referenced with the same URI
4) Ask for human intervention
[others?]

3 is probably just a subset of 2. I'd prefer to see ambiguous 
identifiers as just part of the more general problem of "dirty 
metadata". We can attempt to clean it up - we can chuck it away - or
we can ask a human to help us clean it up. I think Uche is saying
the same thing though probably more elagantly. I also think TBL has been 
consistently making this point:
"The semantic web has to tolerate many kinds of inconsistency."

[1]

> If so, then I completely agree with you.  The only part I am adding is:

> 5) People SHOULD NOT use http: URIs to refer to things like cars
> 6) If they do, there is no guarantee that their metadata will be able to participate in the semantic web.
> 

[...]

IOW we need conventions (or what Uche calls "diplomacy") to help avoid 
misunderstandings. I agree wholeheartedly.

You also seem to accept with your point 6 that we cannot legislate for 
this. As with all forms of communication on the net (indeed - 
communication in general) it behoves us (if we want to be understood) to 
adhere to common convention.

The fact that RDF allows me to say anything about anything doesn't mean 
that anyone else has to listen to me. ;-)

Hasn't this approach allways been a fundamental plank of the web's 
success? I can put up a webpage full of vendor specific HTML extensions 
that break all but a subset of browsers - but I do so knowing that only 
a subset of my potential viewing audience will be able to read it (sure 
people do this in ignorance as well - but we can't .

As to your point 5 - doesn't this approach also allow for the 
possibility that some subset of the SW community may indeed decide to 
use http: URIs to identify cars? Should this really concern us? Or do we 
just accept this as part-and-parcel of belonging to a truly catholic and 
diverse information system?

[1] http://www.w3.org/DesignIssues/RDFnot.html

Cheers,

-- 
Murray Spork
School of Information Systems, Faculty of Information Technology
The Redcone Project
http://redcone.gbst.com/
Queensland University of Technology, Brisbane, Australia
Phone: +61-7-3864-4246
Email: m.spork@qut.edu.au

Received on Monday, 29 April 2002 00:19:33 UTC