22 Jul 2008

See also: IRC log





Sorry for being a bit late.

it's Bill Dutton's office.

there are not really a starting point.

Harry's documents






start with ones close to protocol.

<dbooth> jar: I'd like to see samples of the need for these terms.

<dbooth> harry: Leo Sauerman wants the ability to say "here's how i got this rdf. i followed this 303 and 200".

303 is just a redirection.

<dbooth> ... But he also said if we stick very close to http it may be too close. 303 is just a redirect. Maybe "i followed this linkk from a link header, or an html doc", so maybe some level of abstraction would be useful.

not anymore.

<dbooth> timbl: algorithm i;d use: if you want to generalize http, then find another example of the class first.

timbl: reasonable rule - if you want to introduce a new class.

add a new example.

<jar> (jar and timbl are agreeing! we should have enough eexamples to show that the distinctions are real)

<dbooth> ... show me something that's in the superclass butnot in the subclass.

<dbooth> i agree too

<jar> examples can (should) just go in rdfs:comment.

<dbooth> http://esw.w3.org/topic/AwwswQuestions

We do not have to materialize every distinction in the list.

but we have to have it at least as a textual comment.

<dbooth> should add leo's question to the questions list.

I will see if I can get Leo to write this up and e-mail the list, or join the list.

and discussion.

He is very interested in this.

<dbooth> jar: the important thing about provenance is that what you start with and end up with are different in kind. what conclusions do you want to end up with, at the app level?

<dbooth> ACTION: harry to add leo's question to the question list [recorded in http://www.w3.org/2008/07/22-awwsw-minutes.html#action01]

<trackbot> Created ACTION-8 - Add leo's question to the question list [on Harry Halpin - due 2008-07-29].

<dbooth> ACTION: harry to add textual examples to his terms list [recorded in http://www.w3.org/2008/07/22-awwsw-minutes.html#action02]

<trackbot> Created ACTION-9 - Add textual examples to his terms list [on Harry Halpin - due 2008-07-29].

<dbooth> ACTION: jar to add question about provenance to the questions list [recorded in http://www.w3.org/2008/07/22-awwsw-minutes.html#action03]

<trackbot> Created ACTION-10 - Add question about provenance to the questions list [on Jonathan Rees - due 2008-07-29].

<jar> JAR's intent: meant for "artifacts" to mean things produced by this group

<jar> and for "readings" to be inputs coming from elsewhere... I will fix this on the page

<jar> call it "inputs"? "readings and other inputs?" -- I will try to make this clear.


should we cut it?

specify it to Web agents?

Are all web-agents user-agents?

<timbl> http://esw.w3.org/topic/AwwswListofTerms

state changes.

the resouce has changed.

but i agree, it's very broad.

i am not sure if it should be kept.

Maybe we should generate use-cases first, before walking through the list?

stick to just http

<Stuart> jar, did we give up on: http://sw.neurocommons.org/2008/inforesource2 I thought that we were getting quite close with that.


message which is a superclass of response and request.

so I think maybe payload = message.

is payload more of a packet-level term?

status code.

status code is not an entity.

<timbl> http:statusCdoe is a propert of a n http:response.

response includes metadata.

Hey, I'm just doing what AlanR told me to :)

request is in?

AlanR said "just the terms" first :)

<jar> is 'request' in?

<timbl> http:Request http:Response

<timbl> Both sublcasses og http:Message

<jar> hhalpin: What sends a request? What sends a response?

<timbl> You don't have to model them

<timbl> unless you can conclude things about them

<timbl> Who feeds the people who write the message? <-- out of scope

So, just keep to GET?

<timbl> http:Reply a rdfs:Class ; <-- in David's

it seems like we want to either get *all* the HTTP methods or *none*

<timbl> it seems to me we need to get all or one or two.

<dbooth> i'm happy sticking with GET, and get that working first

<timbl> GET is the 99% case to start iwth

<jar> refresh http://esw.w3.org/topic/AwwswListofTerms if you like

<jar> TimBL: PROPFIND doesn't fit with web architecture very well

So, GET and HEAD.

Any others?

<timbl> Later, but noit now, PUT

<dbooth> But GET's essentially a superset of HEAD anyway, so I think GET is fine to start with.

<timbl> modelling arbirary HTTP methods would not be so useful IMHO

<jar> JAR suggested that a survey of RDF-in-the-wild (cases where people are using HTTP URIs as subjects of triples, where the URI yields a 200 response) would be instructive.

<timbl> http://esw.w3.org/topic/AwwswJarNotes20080630


We could ask Giovanni or some of the semweb serach engine people for stats on this.

we could each make a use-case.

i was thinking of a "provenance-log"

in RDF.

<dbooth> Use cases: http://esw.w3.org/topic/AwwswQuestions

of all HTTP methods sent by my user agents and the responses.

Use Cases / Questions

<Stuart> So harry, a provenance log wrt to what, ie. what are the items whose provenace you are interested in?

let's say a resource changed in my data-set.

And I'm an e-scientist.

<Stuart> eg. from the earlier discussion are interested in the provenance of an HTTP operation?


<dbooth> harry: A friend was building an ont to help people put students in groups, building it off of a trust ont and foaf, and other onts. Sounded good until the germans changed their ont.

<dbooth> ... I want a log of when they changed their ont. I sent this request on thursday and got a 200 with a particular cardinality restriction, then on friday the ont changed.

<jar> this is very concrete - nice

<dbooth> dbooth: maybe take a checksum when the ont is first used, and then when it;s used later if the checksum doesn't match, then flag a potential problem?

<dbooth> harry: yes

<timbl> Use cases - http://esw.w3.org/topic/AwwswHome ?

<timbl> "aims"

<dbooth> use cases: http://esw.w3.org/topic/AwwswQuestions

<Stuart> Well there are some response headers that can, will or should tell you when the resource was last modified.

checksum is a good analogy for my use case: let's say you have an app. do a check first to see what HTTP requests resulted in something, and when something changes in the stream of requests, please flag user..

<jar> I will move TimBL's paragraph "people would like to store..." paragraph from AwwswHome to AwwswQuestions later, after the meeting.

So here's a use-case: give me a list of links you've followed. Include Link headers, Meta links, hypertext links, RDF links.

<timbl_> "If someone says something how does the other person know what htat perioon said" <-- too vague

<jar> and I'll make AwwswQuestions more prominent as a use case list

<jar> hhalpin: Someone should pick over the use case page carefully.

timbl should have told us this earlier in GRDDL :)

<jar> timbl: Number 1 relates to the "GRDDL graph". One can build a system out of "GRDDL graphs".

<jar> David will go over the use cases page

<timbl_> This just bcaome clear in teh RDFa discussions

I'll add the use-case in RDF-graphs.

RDF-graphs as a "log" for provenance purposes.

2 weeks is fine with me.

I guess the game-plan is then once we get the use-cases.

to go through list of terms/ontologies/rules and refine them till we get consensus.

Sounds like a plan.

<timbl_> 'Number 1 relates to the "GRDDL graph". One can build a system out of "GRDDL graphs"." should be more like "One can' give a single set fo assertios but one can define a function (like GrddlGraph(u)) which relates a uri to a graph which a client WOULD get it it WERE to execute a given protocol/network algo, even though clients are not forced to do that

<dbooth> harry, i personally think the need for specific terms will arise in creating an ont and rules to answer the use cases.

<dbooth> so we may not need to specifically go through a terms list first.

<jar> rrsagent. draft minutes

Summary of Action Items

[NEW] ACTION: harry to add leo's question to the question list [recorded in http://www.w3.org/2008/07/22-awwsw-minutes.html#action01]
[NEW] ACTION: harry to add textual examples to his terms list [recorded in http://www.w3.org/2008/07/22-awwsw-minutes.html#action02]
[NEW] ACTION: jar to add question about provenance to the questions list [recorded in http://www.w3.org/2008/07/22-awwsw-minutes.html#action03]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.133 (CVS log)
$Date: 2008/07/22 14:21:40 $

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

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/URI/resource/
No ScribeNick specified.  Guessing ScribeNick: hhalpin
Inferring Scribes: hhalpin

WARNING: No "Present: ... " found!
Possibly Present: P6 Stuart aaaa aabb awwsw dbooth harry hhalpin hhalpin_ jar joined timbl timbl_ trackbot
You can indicate people for the Present list like this:
        <dbooth> Present: dbooth jonathan mary
        <dbooth> Present+ amy

WARNING: No meeting chair found!
You should specify the meeting chair like this:
<dbooth> Chair: dbooth

Got date from IRC log name: 22 Jul 2008
Guessing minutes URL: http://www.w3.org/2008/07/22-awwsw-minutes.html
People with action items: harry jar

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

[End of scribe.perl diagnostic output]