W3C home > Mailing lists > Public > public-awwsw@w3.org > February 2010

Re: Semantics of 301 [was Re: More comments on draft report]

From: Jonathan Rees <jar@creativecommons.org>
Date: Wed, 17 Feb 2010 19:33:11 -0500
Message-ID: <760bcb2a1002171633x550b072by59fbe980b2b6b1e@mail.gmail.com>
To: Dan Connolly <connolly@w3.org>
Cc: David Booth <david@dbooth.org>, AWWSW TF <public-awwsw@w3.org>
I think we're on the same wavelength. I would take a microtheory to be
an RDF graph, and a context to be some particular model of the graph.

The problem of delimiting the microtheory or RDF graph to consider has
to be out of scope - it's essentially insoluble - or at least
orthogonal to the problem of interpreting HTTP. We should just say
that, *given* a set of exchanges, here's an RDF graph we recommend you
play with. We provide an ontology (with formal and informal pieces),
and say that something nice - or not so nice, if an attacker is
involved - might happen if you choose to interpret the RDF graph in
that way. It's a commutative diagram: the interpretation of the
exchanges factors through RDF.

Then we treat 301 the same way we would treat "have you seen Bob" and
don't get caught up in analyzing which utterance of "Bob" referred to
which person. (Now I'm inclined to go back to saying that 301 means
sameAs.) If we succeed at this - and we should, since it looks to be
simple - then we might proceed to look at harder problems. If we want
to start comparing graphs and their models as Cyc does, fine, but
first let's first figure out what the graphs are that are to be
compared.

-Jonathan

On Wed, Feb 17, 2010 at 6:19 PM, Dan Connolly <connolly@w3.org> wrote:
> On Wed, 2010-02-17 at 13:29 -0500, Jonathan Rees wrote:
> [...]
>> Maybe if you could show what RDF graph you would curate in the case of
>> domain name sale or a 301 that changes different(U,V) from false to
>> true that would help. (I don't understand how you would ever deduce
>> that occurrences of a URI in two exchanges could be converted into the
>> same URI in the graph.) But I would prefer to just say this
>> complication of extracting meaningful information from a situation
>> that is on the surface inconsistent is out of scope. The problem is
>> hard enough already.
>
> Jonathan, are you familiar with context logic, lifting and all that?
>
>  Contexts: A Formalization and Some Applications
>  http://www-formal.stanford.edu/guha/guha-thesis.ps
>
> (my larch notes: http://www.w3.org/XML/9711theory/ContextLogic.lsl )
>
> (a pretty good hypertext document about it:
>  http://www.cyc.com/cycdoc/vocab/context-vocab.html )
>
> I sometimes think of HTTP messages as contexts in that sense.
>
> There are 2 extremes:
>  (1) all HTTP replies are in the same context, and every message
>  from each rational party should be consistent.
>
>  (2) each HTTP message is its own context; you can't really
>  tell whether a URI in one message is to be interpreted in
>  the same way as that URI in another message unless you have
>  some really good reason to.
>
> This is one of those things where people are really good at
> figuring it out and machines need a lot of help. e.g. if
> I ask you "Have you seen Bob?" you intuit that I don't
> mean "Have you ever seen Bob in your lifetime?" nor
> "Have you seen Bob in the last microsecond?" but rather
> something like "In the domain of discourse where today
> is all the time there is, Have you seen Bob?"
>
> I think it might be useful to infer that if 2 messages
> are causally related (e.g. you followed a link or a redirect
> from one to the other), then they can be interpreted
> in the same context... or something like that. I've
> never really nailed it down.
>
>
>
> --
> Dan Connolly, W3C http://www.w3.org/People/Connolly/
> gpg D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E
>
>
Received on Thursday, 18 February 2010 00:33:44 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 18 February 2010 00:33:45 GMT