Architecture of the World Wide Semantic Web (AWWSW) Teleconference

13 Nov 2007

See also: IRC log


Noah, TimBL, Stuart, PatHayes, JonathanRees, AlanRuttenberg, DBooth
Jonathan Rees (jar)




<jar> Agenda:

<jar> Review context

<jar> Goals & scope

<jar> Ground rules & organization

<jar> just dumped it becuause i forgot how to tell rrsagent

In scope, Out of scope

<scribe> Scribe: dbooth

jar: Talking about how web arch interacts w sem web for science. We came in with some ontological annoyances, and they seem to resonate w timbl's goal of formalism for http. What's the web analog for doing citation? What can you infer from a 303? What does it mean as an RDF assertion? And it goes on from there. How do you know about documment stability.
... timbl suggested a task group to work on these issues.

timbl: careful about scope getting too broad. there were use cases from lifesci community they weren't sure how to implement -- arch wasn't sufficiently clear. Could look at this activity as interpreting the semantics in an ontology, so that we're not wrangling about which english word we're using. would also allow us to write some rules. I thought we could usefully say what 200 means. I'm putting skin in the game because an http transaction does actuall
... y add to the triple store, and it will issue warnings if something looks inconsistent. Alan asked how can you write a program to ask if the arch is wrong. Shouldn't go all the way to defining the arch, but far enough to write the necessary recipes.

Alan: And trying to take the line w LSID and others, and clarify the argument that you can do everything w http.

jar: What are the boundaries? Maybe not this month or next month, but what are issues that are so trivial that nobody should bring up, or should wait until we have more expreience?

Alan: Some conversatoins to go on the table: For a resource, what is a satisfactory representation? Can it be anything? If one represeantiaon is a photo, another perhaps shouldn't be a cartoon, but a lossy photo might be. Others have said you should be able to construct a resource from the rep.
... Another issue: Location independence. What happens when it moves and the community wants to do something about it -- a resolution ontology.
... Some questions about what a document is. There's an old style sense, of a slowly changing piece of work, and a web phenom like news.google.com that don't fit that model but may be called a document.

<timbl> Not to re-do TAG work when we cna reference it

jar: Also, if you have a URI and you get a 200, how do you determine the referent? What's the relation between the rep you got and the resource? If you only have a rep, how do you determine the resource?

Noah: What if I have a resource for the time of day in Boston? My thought would be that the interaction w http 200 does not offer the promise of reconstructing the resource. Rather, to some fidelity you get a rep.

timbl: Some of these questions we haven't really addressed before. Others have more overlap w existing TAG work, but may need to formalize them for lifesci and reexplain.

<Zakim> timbl, you wanted to say that paths forward include: a set of best practices to be foldedinto as a TAG finding; suggestions on missing functionality and possible implementation

<Stuart> given a TAG concensus ... I could#

timbl: There's a need to ask about metadata about a data source. Maybe write that up and liaise w http WG.

alan: another issue is what to say about time and rdf. Keeps coming up.

<Zakim> alanr, you wanted to ask another issue is what to say about time and rdf?

jar: Document metadata issue is coming up in our group. LSID allows you to go from a URI to either a rep or metadata. There are a few people in our group saying "we solved all this a few years ago and don't want to hear about them". It would be nice to have a story for them.

pat: Need to distinguish between problems to solve and need for better exposition.

<alanr> ack

<Zakim> alanr, you wanted to mention issue of imprecision of specification of 303 see also - can we expect RDF?

<Stuart> +1 to what Pat said

alan: Other issue: question of predictability, when we're looking for metadata. When we do a 303, we'd like it to be obvious that RDF should be there with some basic things in it. Perhaps a description of the resource, usage spec. Can we be more precise aobut what should be there?

<jar> i'm chair & have been forgetting to dequeue -- not used to this! sorry

<Zakim> timbl, you wanted to suggest that in scope is the question of showing how anything you can do with LSID is done with HTTP URIs, but that the issue of LSID being unnecessary and

<jar> i see, dbooth is doing acks for me. thanks - I need to learn how to do this

timbl: There's a tag issue and finding saying "just use http". So in scope for this group would be explaining and embelishing how to use http. This may help in the struggle to say why LSID and others are unnecessary.

<jar> we don't need to evangelize, but we do need to know how to answer their questions

alan: Need to give them a good explanation of how they can do what they want in HTTP. Also: how do you know how many representations there are in a resource? Should there be a way? Also: How to figure out if ssomething is a rep versus a resource? Also: Content location headers. Would be nice to have an explanation for them.

timbl: Makes no commitment that things won't change with time, but does have cache expiry date. Regarding time, every now and then people ask about it, and there's this rathole and people think it's broken if it doesn't have time in it. But the arch is that to the first order, it doesn't have time. But expiry headers ... http has its own notion of time.

<Stuart> Re: Content-Location header: see RFC 2616 section 14.14

timbl: But out of scope for us would be new model of time -- one of the many ways -- that's a huge thing. Reasonable for us: Demonstrate existing http system.

dbooth: Should come up w a list of topics to discuss.

timbl: I propose esw wiki.

jar: It's very slow to run.

<scribe> ACTION: timbl to ask SysTeam about moving esw to mediawiki [recorded in http://www.w3.org/2007/11/13-awwsw-minutes.html#action01]

<Stuart> BTW: 2616 speaks a lot about entities which are 'respresentation' like things rather than 'resource' like things.

alan: Easy to move content if using mediawiki.

Noah: Open to public write?

<Zakim> dbooth, you wanted to comment about giving a story to LSID/other folks

dbooth: I don't think we will ever convince the LSID folks. Trying to do so will merely fan the fires.

<jar> ACTION: jar to compile list of candidate topics for this group based on today's meeting [recorded in http://www.w3.org/2007/11/13-awwsw-minutes.html#action02]

dbooth: Another topic sort of mentioned: the need for citation, a constant (fixed) byte string.

timbl: Maybe we make an FAQ w answers?
... From explaining things more, to how do you do specific things, to things that we havent' covered yet.
... Should do issue handling? FAQs could be regarded as issues also.

Stuart: tracker has been helpful for preparing agendas.
... But tracker only assigns actions to a closed group.

timbl: Could have somebody represent the public and take on their action items.

<alanr> big win for tracker is the keeping of the mail stream around an issue. I hope to integrate this into the wiki at some point

Stuart: If somebody new comes it's a real headache.

<alanr> tracker is the least of problems if someone new comes ;-)

jar: FAQ is a good idea, but an ontology would be a separate item.

Stuart: When do we start needing a charter and normal WG behaviors?

Noah: I think we can keep it loose as long as we avoid areas where IP/patent policy would come into play. Ontology is the only worrisome aspect.

<timbl> Open only to people in one or other group?

dbooth: If this group is the intersection of lifesci IG and TAG members, haven't we already agreed to W3C patent policy?

<patH> Please lets not set out to get to recommendation. Let an existing WG do that hassle.

jar: I think of an ontology as an appendix to AWWW. Maybe for version 2.

timbl: I regard this as being part of the TAG work. Suppose we require that anyone involved... maybe a new wiki ito say unless you've signed the IPP you should not contribute. Every wiki has a license, and you've agreed to the patent policy.

alan: Modify the "save page" to add patent policy agreement.

timbl: Need to login w open ID.

alan: Moderation would help also -- not an excessive burden.

<alanr> mediawiki can be set to approve new logins

<scribe> ACTION: timbl to discuss wiki patent policy issues w ralph and folks. [recorded in http://www.w3.org/2007/11/13-awwsw-minutes.html#action03]

<alanr> +1

<alanr> should we have some reading homework?

Stuart: Post minutes to awwsw list?

RESOLUTION: Minutes will be posted to awwsw list

Next Meeting / Homework

<Stuart> Oh yes... need to post agenda's as well.

Alan: All should read jonathan's note: http://sw.neurocommons.org/2007/uri-note/

<timbl> http://www.w3.org/DesignIssues/Generic

<patH> Meeting time might be an issue. I'm sometimes in California, and it would be 6am there...

timbl: I wrote a design issues doc a long time ago.

<jar> ACTION: jar to revise URI note by 11/20 [recorded in http://www.w3.org/2007/11/13-awwsw-minutes.html#action04]

<timbl> http://www.w3.org/2006/gen/ont

timbl: Also, not linked is an ontology.

<alanr> URI Note

<Stuart> rev <40 have broken html - terminally broken for IE, but Mozilla is more forgiving.

timbl: fairly raw, but defines in n3 relationships in the generic resources document

<timbl> Content-Type, Contenet-location

Alan: Familiarize yourself with RFC 2616? Send out a note if there are parts particularly relevant.

<alanr> Tim's ontology of resources

timbl: Content-Type, Contenet-location have come up the most in conversation.

Stuart: 'entity" comes up.

timbl: HTTP spec is not grounded -- no distinction between a person and a web page.

Stuart: Some of the headers are entity headers and some are about the resource.

<timbl> http://www.w3.org/DesignIssues/diagrams/arch/follow.png

<alanr> Design issues re: Generic resources

<alanr> Tim's doodling about follow your nose

Alan: Another topic: How do sparql endpoints fit into this story?

<timbl> Stuart, maybe background fro trans images in in browser prefs?

<timbl> WFM

Stuart: How do third-party accounts of a given thing fit in?

<jar> 2616 resource = "network data object or service"

<patH> That png image has 5 kinds of line in it. ANy guide to intended meanings?

Alan: Difference between usage spec (or URI declaration) and other statements about a resource.

<timbl> Annotation servers, etc

<alanr> Architecture of the World Wide Web, Volume One

<alanr> +1 to social architecture being a huge part of this

timbl: A lot of the questions I heard in the HCLS mtg last week were about "how do we trust that?" What happens when its 404? A lot of them were social questions. So we'll have to write a lot about how to hold a domain name in the public good. A lot of the arch is the arch of organizations rather than computers.
... So we may explain the social side -- technical patterns and social patterns.

<alanr> I'll be there 14th-16th at least

Pat: I'll be in California mid-Dec to early Jan, so 6am Pacific may be inconvenient for me.

<alanr> RFC 2616

<jar> can someone point me to a W3C meeting HOWTO?

<jar> too late? can't rewrite history.

s/or shall i?//

s/too late? can't rewrite history.//

Summary of Action Items

[NEW] ACTION: jar to compile list of candidate topics for this group based on today's meeting [recorded in http://www.w3.org/2007/11/13-awwsw-minutes.html#action02]
[NEW] ACTION: jar to revise URI note by 11/20 [recorded in http://www.w3.org/2007/11/13-awwsw-minutes.html#action04]
[NEW] ACTION: timbl to ask SysTeam about moving esw to mediawiki [recorded in http://www.w3.org/2007/11/13-awwsw-minutes.html#action01]
[NEW] ACTION: timbl to discuss wiki patent policy issues w ralph and folks. [recorded in http://www.w3.org/2007/11/13-awwsw-minutes.html#action03]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.128 (CVS log)
$Date: 2007/11/13 15:44:01 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.128  of Date: 2007/02/23 21:38:13  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/Jonathan/Alan/
Succeeded: s/y add/... y add/
Succeeded: s/ and give them credibility//
Succeeded: s|Homework|Next Meeting / Homework|
Succeeded: s/rrsagent. help//
Succeeded: s/dbooth, the agenda is://
Succeeded: s/oops. Agenda://
Succeeded: s/logger, pointer?//
FAILED: s/i see, dbooth is doing acks for me. thanks - I need to learn how to do this//
Succeeded: s/what is best ack practice?//
Succeeded: s|http://sw.neurocommons.org/2007/uri-note/40||
Succeeded: s|All shold read jonathan's note.|All should read jonathan's note: http://sw.neurocommons.org/2007/uri-note/|
Succeeded: s/rev will be > 40 by next week//
Succeeded: s|The URI will be http://sw.neurocommons.org/2007/uri-note/||
Succeeded: s/ok, dbooth, will you wind this up with zakim et al?//
Succeeded: s/or shall i?//
FAILED: s/or shall i?//
Succeeded: s/<jar> ok//
FAILED: s/too late? can't rewrite history.//
Succeeded: s/sorry about that, i thought you had gone//
Succeeded: s|s/i see, dbooth is doing acks for me. thanks - I need to learn how to do this//||
Found embedded ScribeOptions:  -final


FAILED: s/or shall i?//
FAILED: s/too late? can't rewrite history.//
Found Scribe: dbooth
Inferring ScribeNick: dbooth
Default Present: jar, DBooth, Alan, Stuart, PatH, TimBL, Noah
Present: Noah TimBL Stuart PatHayes JonathanRees AlanRuttenberg DBooth
Regrets: Norm
Got date from IRC log name: 13 Nov 2007
Guessing minutes URL: http://www.w3.org/2007/11/13-awwsw-minutes.html
People with action items: jar timbl

[End of scribe.perl diagnostic output]