Re: Conforming is such sweet sorrow (was: Re: The meaning of "representation" (was: HTTP URIs and authority))

On Nov 24, 2007 9:12 PM, Pat Hayes <phayes@ihmc.us> wrote:
> :-) Well, I wrote that a while back, and I'd like
> to revise it in the light of HTTP-Range-14, but I
> can't. Ironically, now that I see the logic (did
> I say that??)

Yes you did :) I see an even different irony here: the fact that you
(the author) have been convinced that indeed that page breaks web
architecture and yet the solutions are more expensive than letting it
be.  Furthermore, additional irony in that *I* feel that document
doesn't break anything and presents an important consenting opinion
very clearly (in my opinion).

> behind the HTTP-Range-14 decision,
> and now that I try to be a good citizen, I run
> into practical problems of conforming to TAG
> decisions. How can I arrange for the appropriate
> redirection response to be given? The TAG gives
> me helpful advice, with a variety of scripts to
> use, but all of these require me to modify the
> special file ".htaccess". And I'm not allowed to
> do that: I don't have the permissions.  Moreover,
> my systems administrator, who is under most
> conditions the most helpful and accommodating
> man, flatly refuses to give me or anyone else
> write permissions for that file, apparently
> because to allow access to that file would
> immediately make our firewall protections
> compromised to hacker attacks. So I can't
> possibly follow the TAG advice, and cannot
> conform to HTTP-Range-14.

I really would like to see/understand the empirical evidence that
indicates the 'harm' that URL does in its current state (perhaps I can
dig through some older httpRange-14 threads to find that).  Has any
laptop suffered a shortage in its circuit that led to a civil suite
due to liability etc..,

On a different (perhaps more constructive note), Pat, I think you can
achieve your initial intent by using a standard that embeds RDF
directly into the HTML such that the RDF captures the declarations you
have recorded in natural english.  I've taken the liberty of running
with this idea for your sake, because I think this issue is crucial :)

First, you are in luck because that page is well-formed XML (already
one step further from the format purgatory of tag soup).  Second, all
you would need to change on that page is what is included as a patch
at the bottom of this email (you can go a few steps further to capture
the additional assertions made on that page if you have an appropriate
vocabulary in mind)

I used RDFa with a GRDDL profile to embed the following assertions
(such that a GRDDL-aware agent or an RDFa processor can extract the
RDF).  These were serialized (into N3) using GRDDL.py on a local copy
of the patched file:

@prefix foaf: <http://xmlns.com/foaf/0.1/>.
@prefix pat: <http://www.ihmc.us/users/phayes/>.
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>.

 pat:PatHayes a foaf:Person;
     foaf:birthday "21 August 1944";
     foaf:depiction <http://www.ihmc.us/_images/people/large/phayes.jpg>;
     foaf:name "Pat Hayes".

Now, I have some questions about the hypothetical scenario where Pat
makes this change.  Mostly questions about the "RDF semantics" of such
a page.  Maybe the situation is much simpler than what I envision and
going with a concrete example will make this evident.  If you believe
that the "embedded" RDF is in error WRT httpRange-14, where would that
logical 'inconsistency' be triggered and how would it it have reached
the intelligent agent (i.e., one at least capable of deductive and/or
model theoretic inference)? Neither RDFa (as written) nor GRDDL add
any triples other than those explicit in the content.  It is assumed
that the source document (the message payload) indicates the authors
intent (infallibly).

I personally would like to use a single URI which denotes me
(independent of whether or not the same URI is dereferencable) and
(optionally) resolves to information about me for both a human (web
page) and a machine (RD*,OWL) to 'read.'  HTML fragments in RDF are
nice because the human can get information "in context" although the
machine gets the same information from the URI.  However, as we know,
these can be problematic.  Canonical URIs (those without fragments)
can use 30*s to safely redirect to ancillary information, but when we
want to directly return information, what KR pipeline should an
intelligent agent use to setup a context for inference? All of the
caveats you have to contend with when using an HTTP URI have made me
simply use tag scheme identifiers (or any URI scheme which does not
have a transport protocol - or at least one which does not limit your
toolkit).  It shouldn't be this way, this is why I'm very interested
in why the very simple patterns like these are broken.

> I feel as though I am
> in the position of being told that I must use
> radioactive polonium, but am also told that to
> have this would be so dangerous that I cannot
> possibly have it.
>
> I wonder if others are in a similar bind?

I am actually.  In several separate situations lately, I have found
myself suggesting (and adopting) the use of non-dereferencable schemes
due to issues with control over authoritative web space for an HTTP
URI (or set of URIs).  In some cases it is control over the context
for inference (is httpRange-14 a universal truth in any knowledge
representation that relied on the AWWW?) in other cases it is control
over public network address space (web space).

I'll give you an example.  I recently have been trying to write a
transformation from SNOMED-CT to OWL.  It is an incredibly detailed
ontology of a particular scientific reality (clinical medicine).  The
first roadblock is picking a URI naming convention for the target
vocabulary.  The terminology is owned by the International Health
Terminology Standards Development
Organisation (IHTSDO(R)).  Their network contact info is
www.ihtsdo.org / support@ihtsdo.org.  I don't have control over their
"public" network space, but I can use their email in an info: or tag:
based URI convention.  It is very trivial, so I settled on this one:

tag:info@ihtsdo.org,2007-07-31:SNOMED-CT#

The nice thing about using a non-dereferencable scheme is that you
aren't confused about which class the referent is a member of (it is
completely specified by the RDF content).  In addition (I would like
to believe and would be appalled if this were not the case) I should
not be held liable for violation of any copyright associated with
ownership of the terminology: the things i'm doing with URIs in that
scheme are not visible to anyone else, but the names still have
powerful meaning (they are 'symbols').  Let us say, I *did* have
control over some web space.  What caveats should I be aware of when
putting up such information that was meant for consumption by an
intelligent agent?

-- Chimezie

------------------------------------------------------------------------

--- /home/chimezie/hayes-original.html  2007-11-25 09:11:24.000000000 -0500
+++ /home/chimezie/hayes.html   2007-11-25 09:10:55.000000000 -0500
@@ -1,20 +1,24 @@
 <?xml version="1.0" encoding="iso-8859-1"?>
-<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
-<html xmlns="http://www.w3.org/1999/xhtml">
-<head>
-<title>Pat Hayes</title>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML+RDFa 1.0//EN"
+    "http://www.w3.org/MarkUp/DTD/xhtml-rdfa-1.dtd">
+<html
+  xmlns="http://www.w3.org/1999/xhtml"
+  xmlns:foaf='http://xmlns.com/foaf/0.1/'>
+<head about="http://www.ihmc.us/users/phayes/PatHayes"
profile="http://ns.inria.fr/grddl/rdfa/" instanceof="foaf:Person">
+<title property='foaf:name'>Pat Hayes</title>
 <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />
 </head>

-<body>
+<body about="http://www.ihmc.us/users/phayes/PatHayes">
 <p>Be it known to all agents that the purpose of this document is to
determine the referent of the URI <a
 href="http://www.ihmc.us/users/phayes/PatHayes">http://www.ihmc.us/users/phayes/PatHayes</a>.</p>
 <p> This document hereby declares, establishes and records the fact
that the URI
   <a href="http://www.ihmc.us/users/phayes/PatHayes">http://www.ihmc.us/users/phayes/PatHayes</a>
is owned by Pat Hayes and is intended by Pat Hayes
 to rigidly denote himself, Patrick John Hayes, commonly known as Pat
Hayes, US resident alien and UK citizen, born in the town of Newent,
-Gloucestershire, UK, on 21 August 1944, the only son of Alexander and
Betty Hayes and brother of Mary Ellen Hayes, of Canton, Cardiff: the
legal
-person and living breathing human being whose US social security
number is the decimal rendering of the hexadecimal 21D930EA and whose
facial
-features can be inspected in Brobdingnagian detail by clicking <a
href="http://www.ihmc.us/_images/people/large/phayes.jpg">here</a>.
+Gloucestershire, UK, on <span property='foaf:birthday'>21 August
1944</span>, the only son of Alexander and Betty Hayes and brother of
Mary Ellen Hayes, of Canton,
+Cardiff: the legal person and living breathing human being whose US
social security number is the decimal rendering of the hexadecimal
21D930EA and whose facial
+features can be inspected in Brobdingnagian detail by clicking
+<a rel="foaf:depiction"
href="http://www.ihmc.us/_images/people/large/phayes.jpg">here</a>.

 </p>
 <p>Any interpretation under which the URI <a
href="http://www.ihmc.us/users/phayes/PatHayes">http://www.ihmc.us/users/phayes/PatHayes</a>

Received on Sunday, 25 November 2007 15:00:15 UTC