See also: IRC log
<jar> zakim's clock is fast.
<jar> by 1-2 minutes
<jar> pinging alanr
<jar> alanr: "Few min. Parking"
hi
zakim who is here?
<jar> Look for subject: AWWSW Telecon Tuesday 2011-05-10
<dbooth> jar: In the last few months I've been focusing on how to prevent the train wreck between CC and linked data.
<jar> does the URI refer to the IR at that URI, or to something else?
jimendo
<jar> e.g. another IR, or a toucan
<dbooth> alanr: i.e., whether the URI refers to the page at the URI or something else? this comes up with CC licenses.
eg. music IR at http://www.jamendo.com/en/
<dbooth> jar: This also is related to Ian Davis's desire to ditch the httpRange-14 rule.
<jar> also Ian Davis and Harry H's desire to ditch httpRange-14
<jar> <http://www.jamendo.com/en/album/78807>
<jar> xhv:license <http://creativecommons.org/licenses/by-nc-sa/3.0/>.
ping
ping
<jar> does the URI refer to the IR at that URI, or to something else?
<dbooth> alanr: And in the specific case of a music file, the IR at that publish URI is not a music file.
"a page about an album"
<dbooth> jar: There are two IRs: the landing page and the music.
the IR at <http://www.jamendo.com/en/album/78807> is a "landing page", and it is a page about an album.
<dbooth> jar: Clearly they do not mean to license the landing page.
that URI is a subject of a license assertion
but clearly the license assertion is about the music
<dbooth> jar: CC has a page that tells people how to do this correctly, but jamendo didn't do it right.
http://www.jamendo.com/en/download/album/78807
that's the download URI on the landing page
<jar> http://wiki.creativecommons.org/Web_Statement
that's the IR?
<jar> I get an error
alan asks David if he agrees this is a problem
David says: example of ambiguity. one kind of ambiguity of reference. We can't get rid of ambiguity. Arch works fine.
practical problem: specific applications are counting on one way of expressing things
this one is not following the rules that are expected
Jar: what rule is it not following
dbooth: one class of applications will interpret the license as applying to the landing page
(class of applications that work the way jonathan documented)
but also could be others who interpret it differently?
dbooth: not clear there is this second class
for people in that case the URI indirectly identifies the music. How do they know that's what it identifies? By human means. But don't see an automated way that they could know that.
dbooth: license would stand up in court - intent was clear but is problem for automation
goal is that intent should be clear by automation
are there ways to make this clear?
Harry suggests using IRW ontology - add more metadata that says the URI in question means something particular
<dbooth> dbooth: The problem is how to say that some other statements should be ignored.
<jar> (1) use a different URI - that's what CC says (e.g. urn:sha1: )
<dbooth> ... i.e., statements made implicitly by an HTTP 200 OK response code.
<jar> http://wiki.creativecommons.org/Web_Statement
<jar> (2) use landing page URI, but add metadata to make it clear that this URI refers to the music, not to the landing page.
<jar> (2) is what Harry proposed (IRW)
<jar> (2) scoped to RDF
<dbooth> dbooth: the URI could refer to the combination of the landing page and the music, and then the license would apply to that combination.
<dbooth> ... so if owl:sameAs were used between the music URI and the landing page URI, this may be a solution.
<jar> no
<jar> because how would anyone know that the music URI refers to the music?
<jar> same problem - figuring out what the music URI refers to, and what the landing page URI refers to
alanr: asks the question what the difference between saying "this URI refers to the music" versus. URI sameAs musicURI
trying to clarify (2)
<dbooth> ... an assertion could say that the thing is :Music .
one answer is (2) is one way but sameAs is two-way
rephrasing: If you make a statement that "URI refers to the music" this is one-way.
one-way = unidirectional (synonym in alan jargon)
A sameAs B <=> B sameAs A
sameAs is statement about resource, not about URI
<urn:sha1:MSMBC5VEUDLTC26UT5W7GZBAKZHCY2MD> :license <http://creativecommons.org/licenses/by/3.0/> .
CC example of musicURI is urn:sha1...
<jar> <urn:sha1:x> foaf:sha1 x.
http://en.wikipedia.org/wiki/Magnet_URI_scheme
<dbooth> <http://www.jamendo.com/en/album/78807> owl:sameAs
<dbooth> <http://www.jamendo.com/en/album/78807> cc:license <foo> .
<http://www.jamendo.com/en/album/78807> a :music
<dbooth> dbooth: And as long as classes are not disjoint, then it works.
this help because now you can start looking for something is music
jar: doesn't say which music
but if there is additional metadata (on either URI)
<dbooth> <http://www.jamendo.com/en/album/78807> :composer "Johann sebastian Bach" .
<jar> sure, with enough metadata you can identify what you want to refer to. e.g. sha1
<http://www.jamendo.com/en/album/78807> creator bach
<dbooth> dbooth: So it seems to me that the owl:sameAs solution works, provided that the consuming apps can live with the ambiguity it creates between the landing page and the music.
<dbooth> alanr: People are looking for a follow-your-nose process that will yield the right answer.
<jar> we're talking about apps like ccmixter that need to know what is licenced and how, in order to be correct
<dbooth> <http://www.jamendo.com/en/album/78807> cc:license <foo> .
<jar> it's xhtml:license
<dbooth> <http://www.jamendo.com/en/album/78807> :onWebAt "http://www.jamendo.com/en/album/78807" .
<jar> :onWebAt
<dbooth> per http://www.w3.org/2001/tag/awwsw/ir/latest/
<jar> then we're hosed. the IR on the web at that URI is the landing page
agreed
<jar> in this case the sameas would create an inconsistency
[a landingpage] sameas [a music]
<jar> representation is authorized for uri
an authorized representation from a landingpage(uri)...
would be one that you get by getting that URI
representations for the music are not authorized for the landing page and vice versa
<dbooth> { ?landingPage :onWebAt ?landingPageUri. } => { ?rep ... }
<jar> when someone asks for the zip file, they have to get the zip file
<dbooth> <http://www.w3.org/2001/tag/awwsw/ir/latest/> or "http://www.w3.org/2001/tag/awwsw/ir/latest/"
<dbooth> ?
<http://www.jamendo.com/en/album/78807>
an intended (authorized) representation would be e.g. html encoded human readable information
<jar> assuming <x> :onWebAt "x"
(assuming <http://www.jamendo.com/en/album/78807> onWebAt "http://www.jamendo.com/en/album/78807"
<dbooth> <http://www.jamendo.com/en/album/78807> :hasAuthorizedRepresentation "Content-Type: HTML, ... " .
<http://www.jamendo.com/en/album/78807.zip>
intended representations are encoded digitized music
<dbooth> <http://www.jamendo.com/en/album/78807> :onWebAt "http://www.jamendo.com/en/album/78807" .
<http://www.jamendo.com/en/album/78807.zip> onWebAt "http://www.jamendo.com/en/album/78807.zip"
<dbooth> <http://www.jamendo.com/en/album/78807.zip> dc:author "Johan Sebastian Bach" .
<http://www.jamendo.com/en/album/78807.zip> sameAs <http://www.jamendo.com/en/album/78807.zip>
belay that. let me try again
<http://www.jamendo.com/en/album/78807.zip> sameAs <http://www.jamendo.com/en/album/78807>
<jar> the landing page does NOT have any ZIP authorized representations, by definition of :onWebAt
<dbooth> <http://www.jamendo.com/en/album/78807.zip> :hasAuthorizedRespresentation "Content-Type: Music..."
not by onWebAt, by what we agreed was an authorized/intended epresentation
<jar> sameas leads to R authorized and R *not* authorized
there is no correct interpretation in which we don't have a problem. David do you agree?
<jar> back in 2 minutes
<dbooth> dbooth: Well, it depends on what you mean by "correct".
there no not inconsistent interpretation in which we don't have a problem. David do you agree?
<dbooth> If there are no disjointness assertions, then there *are* satisfying interpretations.
(i.e. correct = (at least) not inconsistent)
<dbooth> dbooth: This may work just fine for applications that do not need to distinguish between the landing page and the music. But it won't work for apps that need to distinguish between them (and which effectively use disjointness assertions).
<jar> there are always going to be satisfying interpretations that aren't correct
<dbooth> jar: There are going to be satisfying interpretations that are not "correct", i.e., do not agree with the real world.
<jar> that just means that we've failed to articulate some axioms
an application that is classifying assets:
making a classification into one of 3 categories
1) intellectual property
2) real estate
3) vehicles
both <http://www.jamendo.com/en/album/78807.zip> and <http://www.jamendo.com/en/album/78807> classify under (1 intellectual property)
so: it's ok to make the sameAs statement
(according to dave)
<jar> db: assuming there is metadata that says that one of the URIs names intellectual property
dbooth says: it is not possible to serve all the application
<dbooth> dbooth: i.e., it is impossible to avoid all ambiguity of reference.
without the sameAs the license attribution application doesn't work
with the sameAs it also doesn't work, because of inconsistency
I'm a publisher and I'm trying to make something that works for both the music spider and the classifier
music spider = license attributor
dbooth: claim sameAs works for license attributor
<dbooth> dbooth: I think the music spider app is different from the license attributor.
proposal: general advise that landing pages should have sameAs to resource would be a bad idea
<dbooth> dbooth: It would be bad only because there are important existing apps (e.g., music spider downloader) that would fail.
what worries me with david's "applications rule"
philosophy is that it seems it encourages building to the unspecified. Namely it places more importance on the computed possible interpretations than on the intended ones (even given evidence of intention)
<dbooth> dbooth: my "applications rule" is that what matters is not whether the assertions are correct statements about the real world, but whether apps work using them.
<dbooth> dbooth: What you're describing is what I call the resource identity guessing game.
<dbooth> dbooth: Yes, my "applications rule" *does* place more importance on the computed possible interpretations than on the "intended" interpretations. The reason for this is that the "intended" interpretations are only known by the publisher. They are not knowable by the users or consumers of a URI.
This is scribe.perl Revision: 1.135 of Date: 2009/03/02 03:52:20 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/goal:/goal is that/ Succeeded: s/hasUri/onWebAt/ Succeeded: s/and/an/ Succeeded: s/is int/names int/ Succeeded: s/ it / my "applications rule" / No ScribeNick specified. Guessing ScribeNick: alanr Inferring Scribes: alanr Present: David_Booth Jonathan_Rees Alan_Ruttenberg Got date from IRC log name: 10 May 2011 Guessing minutes URL: http://www.w3.org/2011/05/10-awwsw-minutes.html People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]