W3C home > Mailing lists > Public > www-rdf-interest@w3.org > April 2002

RE: Explicit Disambiguation Via RDF bNodes, more Process

From: Joshua Allen <joshuaa@microsoft.com>
Date: Sat, 27 Apr 2002 01:00:13 -0700
Message-ID: <4F4182C71C1FDD4BA0937A7EB7B8B4C104F05EC2@red-msg-08.redmond.corp.microsoft.com>
To: "Danny Ayers" <danny666@virgilio.it>, "Sandro Hawke" <sandro@w3.org>, <www-rdf-interest@w3.org>
> to (URI.person, place) and (URI.website, string)
> and someones states that http://www.markbaker.ca/index.html  livesIn
> and someone states that http://www.markbaker.ca/index.html
> Error404 then I can reasonably ask
> Where does  http://www.markbaker.ca/index.html  live?
> and expect the answer 'Miami'.

Is it correct to say that you are suggesting that predicates be typed?
What about a predicate which can apply to more than one type?

In any case, I think your example is broken.  You can ask for the
livesIn value for the URI and get back the value that was stored without
ever having to know that the livesIn and respondsToGET were not making
assertions about the same "thing".  It would be perfectly valid for them
to be the exact same "thing", and you would still get accurate result.

But I assume you probably meant an example like:
dc:Creator (URI.website, URI.person)
and the assertion:
http://www.markbaker.ca/ dc:Creator http://www.markbaker.ca/

Where the fact that dc:Creator was specified to only be able to make
"person" assertions about "websites" is the disambiguating factor.

Is that correct?

> meaning of URIs, doesn't that suggest at least that it is possible?
> thus probable 'in the wild')

Not really, most people want to make their stuff work, and not to prove
a point.

> Sorry, I've gone back through the last dozen or so of your posts and I
> can't find the problem - could you please repeat it. Here's a question

The problem was included in the post.  

>>"There is only one very simple problem being solved here.  If I want
>>allow millions of people to independently share their assertions and
>>that they are talking about the same "thing", how do I *identify* a 

The answer to this is fundamental to the core of a semantic web.

> If I am to use this information in my inference system, how exactly do
> know that the first of these identifies the person, and the second the
> page? (and yes, I would like you to spell it out)

Well, the basic idea is that the "thing" being described; or the "noun",
is what should be as precise as possible.  If Microsoft reports good
results, instead of saying "www.microsoft.com made lots of money this
quarter", I would say "The company that owns www.microsoft.com made lots
of money this quarter".  (If I insisted on using the website URL in my
statement, that is).

Or, step-by-step, the logic goes like this:
a) http://www.microsoft.com refers to a document/website, since that is
what the http:// scheme says.
b) I can't talk about the website when I mean the company, since the
website alone actually *lost* money, and they are two distinctly
different "things".
c) I could just talk about the website anyway, and hope that people
guess that I *really* mean the company, since I don't know the company's
d) Or I could find a way to talk about the company without having to
know it's URI, by relating it to the URI that I *do* know.

Since I am a concerned and civic person, I take approach "d".

I could take two assertions:

http://www.microsoft.com/ isOwnedBy x
x hadIncome lots

Since I know that the *object* of the first assertion is the same
"thing" as the *subject* of the second assertion, it really doesn't
matter what URI they have, so long as it is identical and unique (in
other words, an anonymous node in N3 or RDF).

And in fact, I could simply merge them into one assertion:

(http://www.microsoft.com/ isOwnedBy) hadIncome lots

"The entity which http://www.microsoft.com/ isOwnedBy hadIncome lots"

("The entity" is the anonymous node).
Received on Saturday, 27 April 2002 04:00:27 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:53 GMT