Draft minutes from 2008-04-29 meeting

http://www.w3.org/2008/04/29-awwsw-minutes.html

and below in plain text.

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

   [1]W3C

      [1] http://www.w3.org/

                               - DRAFT -

                                 AWWSW

29 Apr 2008

   See also: [2]IRC log

      [2] http://www.w3.org/2008/04/29-awwsw-irc

Attendees

   Present
          Noah, Stuart, David, Jonathan, Alan

   Regrets
   Chair
          Jonathan Rees (jar)

   Scribe
          dbooth

Contents

     * [3]Topics
         1. [4]Banter before the meeting
         2. [5]Check in on our direction
     * [6]Summary of Action Items
     _________________________________________________________



Banter before the meeting

   Alan: I gave an intro to SemWeb talk. I'll post the slides URL.
   ... We have a an OWL profile for rules. Check the draft.

   <inserted> David, here is the presentation
   [7]http://sw.neurocommons.org/presentations/HarnessingSemanticWeb.pd
   f

      [7] http://sw.neurocommons.org/presentations/HarnessingSemanticWeb.pdf

   dbooth: How it relates to RIF? Alan: Don't know.

   Alan: please review the OWL drafts!

Check in on our direction

   jar: I see a bunch of people with different concerns. Tim is
   concerned with accessibility. Conneg is about making it accessible,
   so instead of linking to a particular langauge you link to a generic
   resource and conneg gives them the appropriate language. So when Tim
   talks about things "breaking the web", my interp is that that's
   because he's worried that the user experience will be degraded.

   noah: i think he has other concerns too.

   jar: there's a set of concerns there that have mainly to do with
   browsing -- classic web arch concerns. There are SemWeb concerns
   that are a different place on the landscape w diff requierments. Eg
   for repeatability of experiement you really do want to refer
   specifically to the XML version of a doc.

   I believe timbl would advodate a uri for both the generic resource
   and each specific resource

   jar: There's different set of concerns where you want to nail things
   down that appear to be at odds w arch concerns in which you want
   things to be abstract. There';s some tension there, but plenty of
   space to deal with both concerns.
   ... i've heard Alan talk about info entities and feeling frustrated
   that there's not a lot of attention paid to it.

   <Stuart> FWIW, I think that dbooth is correct wrt to TimBL separate
   URI for generic and its variants...

   jar: So the q on the agenda: do we want to characterize what's an
   allowable 200 response? I want to check it with people. Is this
   reasonable to do?

   noah: Some of my comment may be because i'm not from the SemWeb
   community. When I see this discussion going on, there's a complexity
   to it. Theres's a level to which i send you this on the wire, you
   send me that ... we're not looking at the most subtle parts of http
   yet (caching, etc). We're looking at the status code 200, media
   type, body. The level of subtlety that we're getting into w
   ValueClouds makes me say "we may need this, but it's daunting and of

   f putting".

   scribe: off putting".

   Stuart: have you seen jar's most recent diagram?

   <Zakim> dbooth, you wanted to say yes

   dbooth: yes, i think we need a new working def of IR, and i've
   proposed a very simple one: a function from time and requests to
   representations.

   <noah> Hmm. I thought semantics of HTTP was about right.

   jar: maybe we should review the aims of this project

   <noah> I'm not sure I want to pull on that bit of string and unravel
   the semantics of the whole world to which that connects

   <jar> http semantics is good, but only a means to an end. why would
   we worry about it?

   stuart: i'm not overly motivated to pick a new def. I'm not overly
   dissatisfied with the one we have at the moement. I acknowledge it
   has some flaws, but any def we have people will criticize.

   <noah> I agree with Stuart. The definition of IR we have seems to
   work better for me than it does for many others on this call.

   jar: It does actually matter when we're giving advice to people what
   to say about a 200 responses. But if its' one of these borderline
   cases that keeps coming up then we can't say. But the other issue is
   provenance and repeatability: you want to make statements that are
   clear about documents, versions of documents, etc. It's not always
   clear what to say about them, and when you do start saying things
   about them you collide with the current AWWW def of IR.

   <inserted> jar's latest diagram:
   [8]http://sw.neurocommons.org/2008/inforesource2.pdf

      [8] http://sw.neurocommons.org/2008/inforesource2.pdf

   <inserted> jar's latest diagram:
   [9]http://sw.neurocommons.org/2008/inforesource2.pdf

      [9] http://sw.neurocommons.org/2008/inforesource2.pdf

   Stuart: how does alan feal about the diagram?

   Alan: neutral. it doesn't really say anything.
   ... It doesn't seems like we're saying much about the SW, but
   there's new stuff happening -- linked data, etc. It resembles
   working w legacy software -- trying to shove something in an old
   design.

   stuart: what sort of things do y9ou want to conclude from a 200?

   alan: difference between a patient medical record and an entrez gen
   record, because one is updated every time info is added.

   stuart: You'd need a vocab for articulating those features.

   alan: Yes, and that vocab would have to go beyond saying
   "FixedResource", etc.

   +1 to alan's comment

   stuart: would a 200 with a link header help?

   alan: yes, it would allow us to start the conversation. we could
   then ask what the description should have. for docs on the web we
   could talke about what kinds of docs they should be, dynamic pages,
   etc. I was going to suggest: let's have an oracle that says: "bless
   you, you can now start the conversation. You have a description, and
   a URI of the thing, and you can talk about them."
   ... Now we can go back and say "how can we get to this stage with
   more motivation behind it?"
   ... The link header is a good idea, but i'm anxious to get on to the
   next level of conversation.

   stuart: jar, this started with "what can you conclude from a 200
   response?"

   <noah> +1 To what Stuart is saying about scope

   <noah> The TAG has a separate issue about linking to resource
   descriptions. I didn't think this AWWSW group was focussing mainly
   on that.

   <noah> When to use 200 does seem to me to be in scope.

   jar: i think i gave that as a concrete question. it is one of my
   concerns. this is hard because there are layers. That's part of a
   larger question of how we can use http to support the SW better than
   it does now. Maybe we dont' even need a link header -- just put a
   link in the html doc. It's a compilcated question because it's in
   layers.

   <Zakim> dbooth, you wanted to say we cannot move forward on
   clarifying the semantics and guidelines arounnd 200 responses
   without a workable def of IR

   <Stuart> Alan... if you get a 200 response an a pile of triples...
   is that 'nothing'?

   alan: I've concluded you can't get anything interesting from a 200
   response.

   <Stuart> example please?

   <alanr> What's the difference if I get a 303 and a pile of triples?

   jar: are 200 journal articles IRs? ftrr:IR defines IR as a function,
   and that doesn't sound like a journal article.

   <alanr> indeed

   <Stuart> But with 303 you *don't* get triples... you get another
   reference to something else... and you get to choose whether to
   follow that reference... and if in following that reference you get
   triples... those triples were served wit a 200.

   <Zakim> noah, you wanted to say fuzziness may not always be bad

   jar: for the purpose of awwsw i'd be fine with going w the def of IR
   that david proposed, and move forward.

   noah: I thought this effort was charger w a narrower scope: http as
   a protocol is an important bit of the web. to be able to come up w
   the rdf statements that describe what you can infer is a very useful
   piece of the web. that's somewhat orthogonal to wether it's useful
   for a particular use case. I think it's a usefule piece of
   gatewaying.
   ... also, it certainly is true thta the def of IR in AWWW is not
   crisp enough to give a clean answer to all questions. There are gray
   areas. Is that a fatal problem? That glass is half full. Its' very
   clear that a human does not meet the def. But if i want to convey to
   you certain words in certain order, i think that's an IR. a
   reasonable person would conclude it can be encoded in bits.
   ... noah: regarding a journal article, when you're deploying on the
   web, you have a degree of control. if what you mean is the text of
   the article, then you have made a decision that what you mean is an
   IR. If you decide that what you meant is something deeper then you
   have decided that it is not an IR.

   <alanr> see
   [10]http://lists.w3.org/Archives/Public/www-tag/2008Apr/0159.html
   starting: "Here's where I think this is going"

     [10] http://lists.w3.org/Archives/Public/www-tag/2008Apr/0159.html

   jar: the problem is that i come up with non boundary cases that
   sound to me like they match that definition, but timbl says they're
   not IRs.

   noah: there is some history here. The TAG was beating their heads
   about this. There was a transition from talking about documents to
   talking about IRs. My example to tim was the abstract tables
   comprising a relational DB. To me a document tends to have a
   beginning a middle and and end, but the relational tables don't have
   that structure. So a rep will have a particular order, but we will
   ignore it. Tim and i continue to disagree on this one.

   jar: when i drew this diagram i was trying to capture what is being
   said about 200 responses. it is almost said in 2616.
   ... there is activity around what 200 means, and some around 303
   redirect. Maybe if we do those two things we're all done because we
   dont' want to take on the def of IR.

   <Zakim> dbooth, you wanted to say that ambiguity and gray area is
   not the primary problem w the current AWWW def of IR - the problem
   is it is wrong

   <inserted> dbooth: ambiguity and gray area is not the primary
   problem w the current AWWW def of IR - the problem is it is wrong

   <noah> My point is that it's a known point of some disagreement
   between me and Tim as to what the word "document" really conveys,
   whether a relational table (in which the order of rows and columns
   is not significant) is best called a document, and whether the range
   of the term IR is the same as the range of the word document.

   <noah> Still, I think that the current AWWW definition of IR is in
   the ballpark, and I mostly find it useful. It does require a bit of
   a sympathetic reading, since it doesn't try to be mathematically
   precise.

   <noah> I have a hard stop now. Need to go. Signing off.

   <scribe> Scribe: dbooth

   s|David, here is the presentation
   [11]http://sw.neurocommons.org/presentations/HarnessingSemanticWeb.p
   df||

     [11] http://sw.neurocommons.org/presentations/HarnessingSemanticWeb.pdf

Summary of Action Items

   [End of minutes]
     _________________________________________________________


    Minutes formatted by David Booth's [12]scribe.perl version 1.133
    ([13]CVS log)
    $Date: 2008/04/29 14:29:17 $
     _________________________________________________________

     [12] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
     [13] http://dev.w3.org/cvsweb/2002/scribe/

Scribe.perl diagnostic output

   [Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.133  of Date: 2008/01/18 18:48:51
Check for newer version at [14]http://dev.w3.org/cvsweb/~checkout~/2002
/scribe/

     [14] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: i/the diagram/jar's latest diagram: [15]http://sw.neurocommo
http://ns.org/2008/inforesource2.pdf
Succeeded: i|the diagram|jar's latest diagram: [16]http://sw.neurocommo
http://ns.org/2008/inforesource2.pdf
FAILED: s|David, here is the presentation  [17]http://sw.neurocommons.o
rg/presentations/HarnessingSemanticWeb.pdf||
Succeeded: i|How it relates|David, here is the presentation  [18]http:/
/http://sw.neurocommons.org/presentations/HarnessingSemanticWeb.pdf
Succeeded: i/My point is that it's/dbooth: ambiguity and gray area is n
ot the primary problem w the current AWWW def of IR - the problem is it
 is wrong
Succeeded: s|David, here is the presentation [19]http://sw.neurocommons
.org/presentations/HarnessingSemanticWeb.pdf||
Found Scribe: dbooth
Inferring ScribeNick: dbooth
Default Present: DBooth, Noah_Mendelsohn, Alan_Ruttenberg, alanr, jar,
Stuart
Present: Noah Stuart David Jonathan Alan
Got date from IRC log name: 29 Apr 2008
Guessing minutes URL: [20]http://www.w3.org/2008/04/29-awwsw-minutes.ht
ml
People with action items:

     [15] http://sw.neurocommons.org/2008/inforesource2.pdf
     [16] http://sw.neurocommons.org/2008/inforesource2.pdf
     [17] http://sw.neurocommons.org/presentations/HarnessingSemanticWeb.pdf
     [18] http://sw.neurocommons.org/presentations/HarnessingSemanticWeb.pdf
     [19] http://sw.neurocommons.org/presentations/HarnessingSemanticWeb.pdf
     [20] http://www.w3.org/2008/04/29-awwsw-minutes.html

   End of [21]scribe.perl diagnostic output]

     [21] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm



David Booth, Ph.D.
HP Software
+1 617 629 8881 office  |  dbooth@hp.com
http://www.hp.com/go/software

Opinions expressed herein are those of the author and do not represent the official views of HP unless explicitly stated otherwise.

Received on Wednesday, 30 April 2008 02:45:29 UTC