SemWeb Best Practices and Deployment

3 Mar 2005


See also: IRC logs day 1, day 2
Photos: Jeremy presenting XSCH, meeting room from front, meeting room from back
Fabien Gandon's summary.


Ralph Swick (W3C), Dan Brickley (W3C), Mike Uschold (Boeing), Andreas Harth (DERI), Chris Welty (IBM Research, Yorktown Heights, NY), Alistair Miles (CCLRC), Steve Pepper (Ontopia), Phil Tetlow (IBM Consulting), Evan Wallace (NIST), Gavin McKenzie (Adobe), Bill McDaniel (Adobe), Fabien Gandon (INRIA), Libby Miller (ASemantics Bristol Office), Noboru Shimizu (Internet Association Japan), Tom Baker (Fraunhofer Gesellschaft), David Wood (University of Maryland), Jeremy Carroll (HP), Valentina Presutti (Univeristy of Bologna), Deb McGuinness (Stanford), Jeff Pan (University of Manchester), Ben Adida (Creative Commons)
Ivan Herman (W3C), Hiroyuki Sato (NTT Japan), Yoshio Fukushige (Matsushita), Marie-Claire Forgue (W3C), Tom Croucher (Sunderland), Balaji Prasad (EDS), Eric Miller (W3C), David Provost, Richard Schwerdtfeger (IBM), Patrick Stickler (Nokia), Charmane Corcoran (Michigan State University), Tim Berners-Lee (W3C), Dan Connolly (W3C)
For selected topics
Steven Pemberton (HTML WG), Mark Birbeck (HTML WG), Michael Sperberg-McQueen (XML Core WG)
Guus, David
Alistair, Jeremy, Andreas, Chris, Ralph, Ivan, Fabien


Jeremythe TAG's reluctance to decide httpRange-14 is becoming a hinderance to the progress of a couple of our task forces

<danbri2> +1

DavidAddison Phillips asked if SWBPD would consider RFC 3066

<RalphS> RFC 3066bis and the Semantic Web [Addison Phillips, 2005-03-02]

Software Engineering Task Force

Phil about to talk about ODA ...

(ontology driven architecture)

phil: at last f2f proposed to send message to IT community who have heard about SW ...
... to say: yes you can use SW to build systems ....
... here is primer on potential, links to technologies, benefits etc. ...
... SE has published early draft of a note ...
... since then have had discussion around content ...
... posted latest version today to mailing list ...
... but has some potential problems - can discuss here today ...
... propose to start there. ...

<RalphS> Latest version of SE Draft [Phil 2004-03-03]

phil: Phil thanks all who helped with the note so far ...

have made substantial changes over last 2-3 days ...

<aharth> http://lists.w3.org/Archives/Public/public-swbp-wg/2005Feb/att-0109/ODA_and_SWSE__Editors_Draft__ver0.7.htm

phil goes through older version of the note ...

describes what has changed in more recent version in response to comments ...

Phil: ... still some issues around style and administratvie content ...
... abstract and intro have been cleaned up, more focussed

MikeU: ... document lacked focus, no clear objective, no target audience, too abstract (referring to old version) ...
... needed to be more meaty ...
... one way forward - specifiy audience ...
... objectives: to get people excited about possibilities for sw technlogies ...
... MDA from UML is powerful framework, we believe it can be augmented with SW technologies ...
... makes it possible to publish & discover ontologies etc...
... then discussed benefits ...
... with automated consistency checking get better quality of software, ...
... maintenance costs reduced because of tie between model & software ...

DavidW: ... aligned with Mike in general ...
... section 3 (meat) did not reflect title ...
... expected to see more focus on ODA ... application of RDF/OWL to this aspect ...
... comments (1) concerns about tone/style (2) concern about direction of note ... latter more important than former ...
... uncomfortable with note moving forward as it stands .... but lots of value in this note ...
... would like more focus on ODA ... find direction to take this interesting note fowd.

Phil: issue of direction ... there is agreement that this thing is valuable. ...
... issue of charter for WG and whether directional note in this context is appropriate ...
... second point: grounding this note in current tecnologies ....
latest version includes section about why OWL is relevant here.

DavidW: would surprise me if some form of ontological approach wasn't used in case tools / case research / other form of MDA in the past ...
... need literatrue survey to find areas where ontlgy approaches were taken ...
then contrast with where OWL makes it better ...
i.e. here is a big win for SW technlgies, here is background to people trying to do this thing in the past, cf. with OWL.

Phil: met to discuss this last night ... have literature refs ...
... after discussion agreed more merit.

ChrisW: was working in this 5 years ago ...
... 30 years of work in using declarative technologies in developing software ...
... what makes it different now is that, although OWL is nothing new wrt KR technologies. ...
... but joining it with the web ... global, accessible, more relevant now, greater chance to succeed now ...
... similar to Java, nothing new but you have global accessibility to a standard ...
... same goes for SW technlogy.

Guus: assuming we get something out there ...
is it convincing and concrete enough to have impact ...
how related to ODM work? ... talk about how to use these things in practise ...
(strawman) give me a reason why we should publish this ?

JeffP: good idea to have as many comments as poss for current draft note ...
... agree with Mike's comments in that current note is too general ...

jjc: I am not part of agreement, do not beleieve document can be rescued in any way ...
dissent because: no clarity about any content, section 3 is the important, but only contains hype ...
... nothing of value in the note ...
... abstract does not relate to content ...
... this wg should not publish this doc ...
... document was circulated too early ...
... where is the value? What is worth doing further work on? Needs convincing.

<danbri> http://lists.w3.org/Archives/Public/public-swbp-wg/2005Mar/att-0033/ODA_and_SWSE__Editors_Draft__ver1.4.htm

<danbri> via http://lists.w3.org/Archives/Public/public-swbp-wg/2005Mar/0033.html

MikeU: maybe could pull out some interesting points from the first draft to put into another doc ...
... potential for note on SW technologies for SE ...
... to jjc - on topics of automated se & MDA of value to write about relation of sw technliges to this?

jjc: not appropriate for this WG.
... lots of other interesting ways that this could be explored ...
... in contrast the OMG docuemnt lacks directional big picture issues, but in terms of useful content it was worthwhile for our user community ...
... i.e. if someone asked me how to use swtechnl for se would direct them to OMG doc. ...

danrbi: scope for middle ground?

jjc: no.

phil: question is who contributed to current document ...
contributors were phil, jeff pan and daniel oberle.

Guus: to evan - should there be a link between the SE draft and the ODM draft?

evan: yes. should this be an 'how to use ODM?' doc ... no.
... there is a good place for a position paper document ... saying there is huge potential, technologies are there, we just need to try it ...
... and ODM is trying to do that ...
... but we do need something that is more esciting than the ODM draft ...

(jjc nods)

jjc: not sure about hypothetical question ... evan's description of doc scope sounded more positive ...
... but a long way from being convinced that appropriate for this WG to write a position paper ...
... but a position paper giving a roadmap doc is valuable ... but what is appropriate forum. ...

Steve: needs to understand what is and is not appropriate for this WG. ...
thought it was about defining best practises and advancing deployment, draw on work already done, on that basis describing best way ...
then how can we further deployment ...
... impression from the SE draft is that it is an exhortation to start doing things, rather than review of what has been done ...
... need to start doing things in practise, then write about doing them (rather than other way around).

DavidW: sensitive to charter, also needs of user community ...
charter says: guidelined that are not based on former practise is out of scope ...
new research is out of scope ...
but in practise have a user community trying to figure stuff out ...

there are real world probs where semweb as whole and business commnuity could benefit from more standard ways of immigrating semweb into se ...

guidance to user community is why we are here ...

agrees that doc is no go in current form ...

but has strong feeling that doc on ontological additions to se practise ... building on 30 years of research ... focussed note on how some of previous approaches could benefit from a semweb approach ...

would be a good note and encourage TF to go there.

ChrisW: only skimmed the doc ...

sees big opportunity, supports idea of this TF ...

se community has lots of momentum into area of overlap with semweb ....

(but maybe doesn't know it)

important time to bridge to that sommunity if we want some influence there ...
... the time is right, the technlgy is right ...

need to take advantage of opportunity to connectg to the community ...

if not they will invent their own technology and we lose a customer.

jjc: there is a case for making some sort of document in this area ...
... maybe easier to connect in timely fashion without going through W3C process ...
... what about other forum e.g. se conference ...
... not opposed to idea that something is publishable ... but still needs to be convinced.

Guus: be happier if note was based on identifying relationship between standards in se community and standards in the web world, abstract from that, high level view for what that could meean in the future ...

MikeU: general note probably not in scope ... but could have more focussed note that is in scope.

Guus: happier with document doc that is about linking standards then adds a 'vision' section to say where we could go with this ...

(instead of just a 'vision' doc)

Evan: not sure what you mean ... because ODM is about linking standards.

guus: could build tools based on ODM that translate UML to OWL ...

so note could talk about this sort of thing then adda v ision section ... ?

<Zakim> aliman_scribe, you wanted to say that it sounds like what is needed is a workshop???

Alistair:I don't know what I'm talking about, but this sounds like networking with people, setting up workshops, outreach
...sounds like out of scope?

<Zakim> jjc, you wanted to record dissent

phil: has been excitiment about workshops ...

will probably happen anyway ...

but why do through W3C? people pay attention to W3C ...

lots of people looking for advice in this area .. look to W3C as authoritative.
but look at it from the outside from a professional who is despserate for guidance who knew that W3C had been playing around with this stuff ...
but then didn't publish anything because of procedure .. looks bad. ...
what would be of benefit is if control of current SE draft is passed to someone else?

danbri: W3C has been changing ... used to prepare things in private fora ... drafts like this not findable buy the public ...

process is evolving to do the work in public ... but there is lack of guidelines for building drafts in public view.

draft has potential but needs more practidcal stuff in addition to vision ...
... there are use cases from e.g. extreme programming ....
... pracitcal examples from collaborative SE .. ?

guus: suggest that a purely visionary document is out of scope for this WG, outside our charter. ...

<danbri> (my ref: pragamatic programmer, http://www.pragmaticprogrammer.com/

could live with a document that contains a visionary section but contains practical links between communities, clearly extablished pragmatics which gives some beef to the vision, acceptable?

Phil: yes.

<danbri> ...chat w/ Dave Thomas, http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/35297

Evan: what do we mean by best practices & deployment? don't agree with jjc, has wider view about the goal of this group.

guus: discuss tomorrow afternnon.
... have clear charter wrt this.

<danbri> (results of hacking w/ Dave Thomas, http://www.w3.org/2001/12/rubyrdf/swdoc/ )

Evan: But have goal to see deploymnent of current tools in new domain ... which is goal of having paper e.g. SE draft ...
... test is we need examples of people already using this in this domain but nobody is doing that.

<danbri> also: http://usefulinc.com/doap is relevant

<tbaker> http://www.w3.org/2003/12/swa/swbpd-charter - the Charter

MikeU: i.e. hard to talk about best practise when nothing is being practiced.

DavidW: uncomfortable with purely visionary work .. but try to find middle ground ...

i.e. there is significant body of former work ... so could put out a doc that says: here's how to take the ontological approach *with OWL* ...

i..e. here is a big win for you by using semweb technlgies ...

Phil: hears that we have agreement to proceed, but need to ground in current technololgies and real world expectations.
... objections?

jjc: some characterisations of the possible path for this document sound ok ... certainly not against some of the characterisations that have been suggested ...

liked Guus's characterisation: grounded in relations between work already done in SE community, work already done in SW commnty, links between ...

<danbri> (maybe we could tweak the taskforce charter to capture whatever this concensus is...?)

Guus: hear consensus to move fwd with this doc in this direction.
... propses action to phil to updated SETF charter accordingly.

<danbri> danbri: (said) maybe we could tweak the taskforce charter to capture whatever this concensus is...?

<em> this is a 'bridge' document between communities... the more study the bridge in terms of concrete connections the more weight this bridge can support in bringing people over and understanding how these communities relate

Phil: thinks doc would benefit from someone else taking charge.
... any volunteers?

jeff agrees to take the lead with the SE draft.

<danbri> (applause for jeff)

ACTION: Phil to update SETF charter in light of new focus for SETF draft note

<danbri> JeffP, I'm moblogging a photo of the room so you can visualize us ;) should show up soon in http://www.foaf-project.org/2004/media/

Vocab Management

Tom: introduces document using slides

<DanC_> (hmm... identify terms with URIs... rather use URIs for terms? URIs like rdf:type don't identify terms; they are terms)

Bleeding edge? means where definitive answers not yet available

<danbri> (DCMI is Dublin Core Metadata Initiative, see http://www.dublincore.org/ )

In third section each issue is treated with two paragraphs

indicating different positions and links to further reading

DCMI documents can fit into this VM note

re other vocabs that are online in a "pre-SemWeb" way...
(many are thesauri; i wonder the relationship to SKOS...)

ACTION: TomB to post URL to his VM TF slides

<DavidW> I like TomB's ideas for a 3rd party endorsement model for vocab extensions

<DanC_> simplest way for DCMI to endorse such a statement is to say it in a document they publish, seems to me

<DavidW> Endorsement is different from original assertion

Shared formalisms -last slide - particularly between foaf dc and skos communities

<DanC_> not necessarily, DavidW

LoC issue to do with endorsement is current concrete problem facing DC community

LoC = Library of Congress

<DanC_> endorsement at the document level is straightforward. Endorsement at the statement level is more tricky.

Tom finishes talk.

<Zakim> danbri, you wanted to mention http://www.w3.org/Talks/2005/0229-jk-rdf-sig/ from SWIF F2F on tuesday (DCMI can just say things re MARC on their site; but can explore digital

DanBri draws attention to talk by José

<Zakim> aliman, you wanted to talk about note scoping

Ralph points out that signing is new work, and hence out of scope for this paper

Alistair: Tom wants us to discuss scoping of current note

<danbri> (also I should've said, just pls take a look thru Jose's slides, if you missed his talk... was only an aside re this current agenda item)

Alistair: however title seems inappropriate e.g. "managing a vocab for SW - review of current practice"

<DanC_> +1 title should say "this document asks more questions than it answers"

Alistair: best practice may not be cuirrent practice

TomB: the middle bit of doc is good practice

Alistair: I've just changed my mind ...

Ralph: I hope this TF will propose best practice

<danbri> (re how we do stuff in FOAF scene, last thing I wrote on this was in http://rdfweb.org/topic/FOAFCommunityProcess )

Ivan: these questions come up a lot, examples of how people approach these questions would be very valuable

Mike: suggest title should be "Managing Vocabs on the SW" not "for"

Jeremy: I think "best practices" means "best current practices"

<Zakim> DanC_, you wanted to offer to fly by some TAG issues that seem VM-related, now that VM moved to today

Alistair: howabout "Managing SW Vocabs"

DanC: couple of TAG issues related: numbers 8 14 35

<danbri> http://www.w3.org/2001/tag/issues.html

<danbri> http://www.w3.org/2001/tag/issues.html#namespaceDocument-8

<danbri> http://www.w3.org/2001/tag/issues.html#httpRange-14

<danbri> http://www.w3.org/2001/tag/issues.html#RDFinXHTML-35

DanC: namespace Document 8
... this is vocab management related ....
... TAG has been discussing RDF Schema's XML Schema, RDDL, HTML docs
... the XML Schema validation service will follow RDDL docs

DavidW: what do you want this TF to do?

DanC: I'm just drawing your attention to these
... httpRange 14
... what is the range of http deref function?
... this is the hash versus slash issue
... RDF in XHTML 35 will be elsewhere on agenda

RalphS: all three of these block deployment of applications and some of our TFs
... I would propose that WG find a TF that is responsible for each of these three
... I suggest we give actions to TF to develop WG position on each of these

JJC: seconded

<danbri> (re -35/xhtml, that is a part of the namespace doc issue)

<DanC_> (oops; I forgot one... "social meaning" has a home in the TAG issues list http://www.w3.org/2001/tag/issues.html?type=1#rdfURIMeaning-39 )

DavidW: xhtml35 is in rdfhtml tf

TomB: I would like to get a short note out quick, and not get hung up on these issues
... going beyond recording current TAG position is something we should do later

DanC: acknowldgeding existence of issue is fine,

Danbri: http-range14 is a bigger block than the others, since namespace docs can be changed without as much disruptive as namespace URIs
... prioritizing
... as namespace owner with URI ending in /
... with limited attention we should work on this issue not namespace docs
... easier to change namespace doc than namespace URI

<DanC_> (you could be more explicit, danbri: it's easier to change a namespace document than to change all the documents that refer to it)

Guus: asking DanBri should we be taking a position and reporting that to TAG?

<danbri> (nice formulation, danc)

Ralph: I would like it to be more explicit, the WG should acknowledge its responsibility to state a position

<danbri> (I don't think VM TF first Working Draft needs to wait for a position on http-range-14)

Ralph: we should not punt this to TAG

<danbri> +1

<jjc> +1

Jeremy: ask for straw poll on httpRange 14

Guus: prefer to have discussion tomorrow

DavidW: issue httpRange14 deferred to tomorrow 12 and 1

TomB: I want to ensure we set milestones for VM note
... March is difficult
... is it reasonable to have first draft by mid-May

Guus: mid May is a bit far away

TomB: still awaiting some input

<DavidW> We are over time for this TF now and need to deal with immediate planning issues. DanC has raised the issue and we will determine a WG consensus on it tomorrow when we have time.

<danbri> (re timing/contribs, I only have time to commit in April... march is taken for EU bids; may is also uncertain)

<DanC_> (er... it's an editor's draft now.)

TomB: if input came mid-April, then we could circulate an editors draft by end-April as 'candidate working draft'

<danbri> (new terminology, but also used in DAWG...)

<DanC_> i.e. a proposal from the editor to the WG to publish as WD

Guus: who in particular are we waiting for? (inputs to VM note)

TomB: pillars are DCMI Foaf skos and relevant TAG issues

DavidW: what is a realistic time? (asking TomB)

TomB: for foaf we need text

<DanC_> DanC: [after FYI re tag issues]. I encourage you, while working on vocabularies, either as a WG (ala SKOS) or individually (ala foaf, ...) to be aware of your approach to these issues and think about whether you'd advise others to do likewise or not.

DanBri: I can't do it in March, but can in April

<danbri> (well, maybe last week of march...)

Alistair: can we do it faster

TomB: I want input from DanBri and Libby, and they are not available in March

Alistair: but the foaf bits are only an hour's work

DanBri: have you factored in procrastination time

Guus: it would be good to have this out soon

Jeremy: let's publish without foaf just a tbd

TomB: draft is currently in Wiki

<danbri> there is http://esw.w3.org/topic/VocabManagementNote

<danbri> (latest?)

<danbri> ah, Tom: I have a version that goes beyond that

<danbri> I see " TASK: DanBri or Libby - Describe W3C usage of the word "namespace""

DanC: I read through this, but I can see it taking it significant time, it's not ready to go

<danbri> a big job in itself

TomB: agreed

Guus: with a midMay schedule for WG vote
... timeline is that we aim for WG vote midMay, 'candidate working draft' two weeks before


Alistair: trying to get input from the group on technical bits
... presents SKOS Core Vocabulary Specification W3C Working Draft In Preparation

Alistair: policy statements: naming (how we form URIs), persistence (URI should stay for a time), change (how URIs change), maintainance (how vocabularies evolve)

<DanC_> (hmm... http://www.w3.org/Consortium/Persistence hasn't really been enacted)

danbri: vocabulary definition should end with a hash

<DanC_> <owl:Ontology rdf:about="">...</> ...

<danbri> aside to report RDFS namespace practice: [[

<danbri> <owl:Ontology

<danbri> rdf:about="http://www.w3.org/2000/01/rdf-schema#"

<danbri> dc:title="The RDF Schema vocabulary (RDFS)"/>

<danbri> ]]

<DanC_> is that # really in there, danbri? that surprises me

<danbri> yes

Alistair: uppercase/lowercase convention for classes and properties

<danbri> alistair, http://lists.w3.org/Archives/Public/public-esw-thes/2005Feb/0054.html is my review comments on this; quite a lot of comment re policy aspects (+ draft text)

danc: points out that the Persistence Policy is a draft

<DavidW> jjc, aliman has raised it as an issue and has an intention of covering it as part of the document review

<Zakim> jjc, you wanted to ask about RDF and OWL vocab management??

ACTION: Ralph to inform the W3C Communications Team that we intend to cite http://www.w3.org/Consortium/Persistence as "W3C URI Persistence Policy"

<DanC_> what change was just agreed by the editor?

ACTION: Alistair to change the wording of the link to "the persistence policy at URL http:"

<RalphS> (some change which I hope doesn't result in a URI-in-your-face)

<DanC_> ew ew ew. please don't do "policy at..." i.e. don't use in-your-face URIs, please.

<RalphS> i.e. I hope the editor takes the intent of the ACTION wording and not the precise letter of that wording

Alistair: re change, three levels of stability: unstable, testing, and stable

<danbri> tom: see 'dcmi namespace policy'

<DanC_> google nominates http://dublincore.org/documents/dcmi-namespace/

<DanC_> Namespace Policy for the Dublin Core Metadata Initiative (DCMI) Date Issued: 2001-10-26

<danbri> (alistair, see 8) Policy Statements in my http://lists.w3.org/Archives/Public/public-esw-thes/2005Feb/0054.html )

em: Is the persistence policy in prose or in machine-readable format? I'd like to see stability statements made in machine-readable form within the schema declarations

<Zakim> danbri, you wanted to try to smmarise my comments from http://lists.w3.org/Archives/Public/public-esw-thes/2005Feb/0054.html

ACTION: Alistair to think about machine-readable change policies

<Zakim> DanC_, you wanted to note http://www.w3.org/2000/10/swap/util/changePolicy.n3 and how I got stumped on TAG issue 41 and ontaria policy

danc: port document focus is thesauri, various tag issues are related to the issues raised in the draft

tbaker: articulate the larger context in which maintainance of terms is embedded

<Zakim> jjc, you wanted to talk about life after WG

ralphs: javascript probably doesn't conform to pubrules; is it important to use in the SKOS Core spec?

<DanC_> (jjc, I'm not sure how OWL got out without a persistence policy in its namespace document... current pubrules prohibit that)

<danbri> [[[

<danbri> I am not personally in a position to make such pledges. Something

<danbri> milder:

<danbri> [[

<danbri> The Working Group is committed to a public, consensus-driven

<danbri> design environment for SKOS, and to this end conducts SKOS-related

<danbri> discussion in public, in particular drawing on feedback from the

alistair: presenting maintenance section that describes the procedure to change skos (consensus...)

<danbri> Semantic Web Interest Group mailing list public-esw-thes@w3.org .

<danbri> ]]

<danbri> ]]]

<tbaker> 1+ the points seem right

alistair: should we leave the four things (naming, persistence,...) in the documents?

ACTION: Alistair to change the links to examples

<DanC_> (I think the lack of established norms is the raison de etre of this BP WG)

<danbri> (+1 danc; I think we're test driving some VocabManagement ideas via this spec too)

<DanC_> (quite)

ralphs: what's the maintainance policy of a Note?

jjc: should we link to the process document (which we got past of it)

danbri: three draft paragraphs to connect w3c processes with skos processes

guus: publish all skos core and skos guide together?

alistair: review status of guide: appraoved for first working draft

<DanC_> (ah! now I know why we're not talking about thesaurus porting very much.)

guus: planning on timeline for three documents?

alistair: 24th to go for all three documents
... third one reviewed once, second draft in review at the moment
... target 17 March for updated versions of all 3 docs for discussion on 24 March

Applications and Demos

guus: what is a good application as convincing application for semantic web?
... business area and non-profit areas

libby: slides, might answer these questions
... weblog has descriptions about applications, using grddl and xslt
... doap vocabulary
... weblog difficult to use
... doap (description of a project) seems to be popular and fits well

<danbri> re DOAP, see http://usefulinc.com/edd/notes/DOAP

libby: better workflow would be that people use doap to describe their projects themselves and TF links them
... what should be the criteria for inclusion?
... presents doap descriptions from the weblog in a facetted browser
... maybe combine DOAP submission with swig mailing list

<Zakim> DanC_, you wanted to say moving DOAP out to the projects sounds great, but let's not let the tools discussion dominate. I think the Nature RSS paper is the best answer to Guus's

mike: there should be a form to create the descriptions

<danbri> (see http://doapy.bonjourlesmouettes.org/doap-a-matic for a form that generates DOAP project descriptions)

danc: good strategy using doap, but tools discussion shouldn't dominate
... rss nature paper best answering guus' question

libby: what should be the policy for inclusion?
... open source downloads and applications available online

guus: criteria from the semantic web challenge (maybe a subset)

<danbri> see http://challenge.semanticweb.org/

<danbri> http://www.informatik.uni-bremen.de/swc/swapplication.html

mike: don't exclude big impact semantic web applications because they're not download-able

<Zakim> danbri, you wanted to be liberal re inclusion

danbri: swc site uses frames

<Zakim> RalphS, you wanted to comment on persistence

ralphs: persistence is of varying importance to different user groups. For me, the most important group is those thinking about adopting SemWeb so listing current work, not stuff that has gone 404, is most useful.

<danbri> (re Frames, see http://www.w3.org/TR/WCAG10-HTML-TECHS/#frames )

<Zakim> pepper, you wanted to ask if TM apps qualify...

steve: do topic map applications qualify?
... ie. omnigator

<Zakim> em, you wanted to suggest a criteria

<DanC_> omnigator... hmm... I can imagine a convincing case for its inclusion.

<danbri> RSS/Nature paper was in D-Lib magazine, DanC. See http://www.dlib.org/dlib/december04/hammond/12hammond.html

?: what about non-english applications?

<BalajiP> Nature RSS paper: http://www.dlib.org/dlib/december04/hammond/12hammond.html

<pepper> well, the omnigator is a bit special, DanC_, because it *does* support RDF

<pepper> (it's also free, in case anyone was wondering)

<Zakim> FabGandon, you wanted to talk about different uses of the list and process

mike: the selection shouldn't be too restrictive

FabGandon: also concerned with the process (review in the group?)

davidw: possible to seperate between open and closed source and profit vs. non-profit
... maintained and not maintained

guus: discuss the process of inclusion tomorrow?

<DanC_> The Role of RSS in Science Publishing

<DanC_> D-Lib Magazine

<DanC_> December 2004

<DanC_> Volume 10 Number 12

<pepper> a tribe of competing 'street' standards :-)

RDF-in-HTML TF discussion (Ben)

ben: xhtml wg is working on rdf/a
... we hope
... we're getting an update today
... re GRDDL we are considering moving towards REC
... is currently a CG Note
... the only one so far existing at W3C
... a bit of a no-mans land
... discussion that having it as WG-based REC would help

guus: what's status of comments on GRDDL?
... what would REC-track take care of?

ben: we need to make sure list of usecases is complete

<danbri> HTML WG arrives

<DanC_> i.e. Steven Pemberton,

stevenp updates us on status

<danbri> stevenp: "we are discussing draft for LC WD of XHTML 2"

Steven: the xhtml2 wd has been updated with the rdf stuff and we're discussing that now

<danbri> WG: http://www.w3.org/MarkUp/Group/

<DanC_> (which of the links atop http://www.w3.org/MarkUp/Group/ is the relevant one?)

Steven: the mapping to triples is in there, though not in the depth taht it is in the rdfa document

<RalphS> [Editor's] draft XHTML 2.0 24 February

<RalphS> (Member restricted)

<Steven> We weren't aware

Steven: rdf/a document needs to be updated

<Steven> but no prob


<ChrisW> what section should we be looking at?

<danbri> meta and rdf is http://www.w3.org/MarkUp/Group/2005/WD-xhtml2-20050224/mod-metaAttributes.html#sec_22.2.

MarkBirbeck: history - decided it would be better to separte the metadta stuff from the html stuff - this is the rdfa document. xhtml2 draft represents the final thinking

<DanC_> (ah... ok, found some examples in 22. XHTML Metainformation Attributes Module )

stevenp: the only change is that only xhtml:about is inherited, everything else has to be declared explicitly

<DanC_> (hmm... "Sorry, Forbidden" at http://www.w3.org/2002/06/xhtml2/ )

MarkB: thought they were finished with rdfa and then issues like making a page it's own foaf page or own rss feed.
... href becomes special to solve this problem

<Zakim> danbri-laptop, you wanted to ask HTML guys about test cases and QA of RDF/A design (RDF syntaxes are hard to test) (when they arrive here...)

danbri: being sure you're right - in rdf wg first one was prose only, rdf core used testcases - do you have time to do this sort of thing?

mark: has tried to use the testcases, but many of them don;t carry over

danbri: the methodology rather...

mark: right, tried to convert rdf document testcases to rdfa and then see if we get to the same n-trioples, but would like some more tests
... using real documents eg foaf, be good to agree on some documents to use

<danbri> (yup i think having real, use-casey examples + their ntriples would be better than using obscure corner cases)

ivan: richard ishida and others have a groupo called international tagset
... these might be close to what your inetrested in

ralph: is thrilled to see a document updated. his expactation was that rdfa wopuld get merged into the working draft and then disappear as a separate document - I think you're saying that it has life as a separate document. priority for him is the working draft not rdf/a
... is the material from the standalone document that we reviewed going in as is or substantially different?

stevenp: not substantially diffrent but b-nodes stuff didn't make it in to this draft

<ChrisW> can someone post the RDF-A url

<benadida> http://www.formsplayer.com/notes/rdf-a.html

<benadida> (rdf/a)

ralph: jeremy implemenbted the oct draft and seemed to end up with many more triples than expected - was that a prose mismatch to number of triples ... etc
... repeating jeremy's tests would be good - have you looked at that?

<danbri> (ralph's point is exactly why I'd like a test case collection...)

steven: think all you have to do is change the namespace. otherwise it should just work

mark: the audience for rdf/a and xhtml2 is quite different
... not indepth explanation of rdf in xhtml2 draft
... point of rdf/a was to have somethign that we could all discuss...
... this should not stop xhtml2 going on its way but steven and I do want to finish rdf/a
... xhtml2 document not sufficient to determine if you get 3 triples or 5

steven disagrees

guus: was your intention to make the drafft less precise, or is this an error

mark: deliberate, for a different audience

steven: suggests that you look at it and feedback to them

ralph: it's important that the syntax is tied to the level of mathematical precision that rdf has
... and that level was in the oct draft
... it needs to be there somewhere

mark: the whole reason for us producing rdf/a was to have these discussions

ralph: his expectation was that rdf/a was a vehicle for discusssion but that it would be reintegrated into the maintream of xhtml2

mark: not necessaily a problem but one motivation for taking it out was so that other languages could use it e.g. svg

Mark: svg has a 'metadata' element, which as far as i can see is completely wasted...

ralph: fully supports that, although there may be objections that its out of scope

guus: we're not trying to bring you more work...and we can probably help

danbri: compromise perhaps: link to the examples and to an exactly equivalent rdf/xml version

ralph volunteers to help

ralph: the taskforce on behalf of the wg should provide some input

ben: maybe usecases, rdfa/xhtml2, rdf/xml could form a document

danc: happy to review it when it's on the TR page

stevenp: not going to release a working draft version
... suspect 2 last calls will be needed anyway

<DanC_> (er... who ended up with the action there?)

<benadida> (I ahve the action)

<DanC_> (tx)

stevenp: one thing that's emerged from discussions today is that saying what meta and rel relationship is to rdf
... link/meta [somethings] are now the same things [something] (sorry

<Steven> http://www.w3.org/MarkUp/Group/2005/WD-xhtml2-20050224/mod-metaAttributes.html#s_metaAttributesmodule

stevenp: for example... need to specify the values for rel and the values for property, e.g. copyright
... e.g. start is only sensible as a rel value and not a property value, but contents could sensibly be both
... just for information that falls out of this approach. not in this draft
... just for information, this is something that falls out of this approach. not in this draft

danbri: which namespace are these in?

stevenp: xhtml2 namespace

danbri: will there be an rdf schema that defines them?

stevenp: we would love help with this

ericm: using rdf/a to declare this stuff would be an excellent testcase, and educational

mark offers to show a real live xhtml2 document

<danbri> (ben, I'll help if needed w/ schema stuff...)

mark: an rss reader needs multiple lists of rss feeds, like OPML, but with more information

<benadida> (dan, sounds great, taking that action)

<benadida> (danbri, that is)

<danbri> (I don't think it's an action yet, but if it's one I can do in April, I'll certainly take it...)

mark: shows an xhtml2 document, with some meta names, a 'nl' navigable links' tag; meta statements in teh body of the document (inheriting from a previous href immediately above it); has an image inline - the document is the metadata

(I didn;t record an action)

<DanC_> MarkB_, would you please mail that example to public-rdf-in-xhtml-tf ?

<benadida> (the action is to provide help to XHTML WG in defining an RDF/OWL schema for the special properties defined in their doc)

[discussion of escaped / unescaped html in meta]

guus: timelines and plans need to be diiscussed - any more qs about rdf/a?

danbri: is it possible to validate that peopel haven;t left out namepace declarations?

<DanC_> (hmm... dunno what relaxng does with qnames in content)

mark not sure.

<Steven> mimasa?

gavin(?): [missed the detail, sorry]...

<jjc> schema validation includes qnames in attributes

<jjc> (XML Schema)

ben: this is great, looking foreward to giving you some testcases

ACTION: DanBri help write an rdf schema for the additional xhtml2 namespace elements

though he may need a telecon and be pinged with the specifics

<Gavin> gavin asked about how to distinguish between metadata on resource referred from parent element href which is inherited versus metadata on parent element

ben: our plan is to cheer you on with rdf/a and fully endorse this for xhtml2; prior htmls endorsing GRDDL, maybe a WG note or a rec
... need to discuss what this would need

<Gavin> gavin understands that if href is inherited as an about, that you'll be able to associated the metadata with the linking element itself (instead of the referenced resource) by doing something like about="" on the meta element

guus: also thinking of writing a very short note about the 2 possible routes for rdf in html, with soem examples
... are you planning anything similar? if so we should coordinate

stevenp: that fits

jeremy: from what I've gathered is the most unwelcome change form the oct doc is the dropping of support for bnodes, so we can;t serialize all rdf graphs
... I think worth highlighting as a signifant change

<benadida> (I had not realized the bnode change either)

stevenp: if a significant problem, review it and discuss how we can get it back in

MarkB: I hadn't realized that bnode support got dropped until Steven mentioned it just now; I will trace back what happened. I think we did have a simple solution...I think we can incorprate it fairly easily

<DanC_> (hmm... something like "pronoun" rather than bnode?)

stevenp: would like to find a way of expressing it that doesn't use rdf technical terms like 'bnodes' - should be something your grandmother would understand

<danbri> re bNodes, FOAF use case I tihnk will need bnode support to capture common FOAF idioms; http://rdfweb.org/mt/foaflog/archives/2003/07/10/12.05.33/

mark: at one point we had xpointer thing, then bnode as an attribute, then object or thing or thingy, need to retrace thought process

guss: feedback on this before last call?

<DanC_> (er... hey... you can't make last call comments to yourself)

<tvraman> raman: virtually here --- rdf:role=observer

stevenp: last call is as soon as we can, so a last call comment makes sense

jeremy: we had a rquest from html wg to help with rdf/a - now you need more help with schema document - which is cool - but would like a better communication process

stevep: weekly 30 min call?

guus: the rdf in html tf calls obvious point of contact

mark: not been clear sometimes if dicsusing something I need to be there for or if you are discussing GRDDL for example

ben: that mailing list rdf in html has been almost all rdf/a. we'd love to have you for those (or part of them)

ACTION: BenA set a time for the RDF-in-XHTML TF telecons

guus: we will send you the minutes of this meeting

danbri: was suprised by the bnode suipport being lost, bnodes very important
... do we expect RDF/A documents to be GRDDLable? and is there anything we need to do
... jeremy and maxf have made xslt...what do we need to do with the schema?
... decide that that's the way to make these work together?

danc: qnames may make this tricky

ACTION: the rdf in html tf to discuss whether GRDDL needs to work on XHTML2 documents

<DanC_> the practicality of using GRDDL on RDF/A documents is impacted by the use of qnames in content

(was that an action?)

<danbri> (jjc, do you have an xslt that can do the qname thing? does it require xslt2?)

<benadida> Ralph suggested naming the various documents so we know which ones we're referring to

ralph: is confused because is rdf/a was standalone, then would just reference tit from xhtml2 document

<jjc> my rdf/a thing was xslt2, but then I've forgotten xslt1

mark: rdf/a not even a working draft so can;t reference it, timelines are wrong.
... ideally it would have been published first and referenced

<RalphS> it appears that we need two names to be able to refer to a standalone module and a module of XHTML2

Mark: there are actually 3 modules of XHTML2

[HTML WG leaving]


GRDDL discussion - future of GRDDL

ben: what need to happen to GRDDl document to have it go in a recommendation direction [?]
... we want feeback form the WG on bringing GRDDL to rec track

danC: the process of note/rec etc is a means tgo an end for danc - get peopel to publish rdf data. rec track makes it easier to justify the time on testcases etc
......q: is this a 'best practice' for the sweb?
... wondering if rec might bring people out of the woodwork
... to get feedback

davidw: personally thinks GRDDL is cool, and like to see it rec track, but not sure if it's in charter

ralph: chater allows for recommendation, specifically for the embedding issue

davidw: withdraws comment

jjc: I need to talk to colleagues about this, can't respond at this time

danbri: comments would be good to hear e.g. from jena devlopers

ben: yes re comments, but we don't have anything else for pre-xhtml2 without it

ralph: this could be addressed by a note

<DavidW> Can someone please show me where in the SWBP Charter we are allowed/expected to solve the RDF embedding question?

ralph: the TF is inclined to move forward with a new version of the document; we need to say in there what plans are

<jjc> The Working Group will, in conjunction with the HTML Working Group, provide a solution for representing RDF metadata within an XHMTL document.

<jjc> 1.2.2

<DavidW> "Produce a Working Group Note on guidelines for transforming an existing representation into an RDF/OWL representation."??

ralph: does the wg share the same opinion as the TF?

guus: what does the TF think it will take in resources to go to rec produce new draft - enough recources?

danc: would like to go around the table and see if it's important to people

jeremy: thinks if GRDDL was rec track it would make a difference to whether to implement it
... guesses not that high priority for jena team

<Zakim> danbri, you wanted to ask (if we want actual discussion of GRDDL detail) whether GRDDL works fine w/ XSLT1 and XSLT2; spec doesn't mention version currently, perhaps we consider

danbri: if its developers vs publishers, he would be hard on the developers
... likes the idea of a big push, e.g for foaf data
... would like to know how hard it is for developers tgo do it, what it entails

gavin: makes me think of blogs...does it make sense to ask member and non-members to see if e.g. rdf comments would be useful?

danc: talked briefly about this - depends if blogs produce xhtml or not
... worth asking them

pepper: still not sure why we can;t just use RDF/A

ben: rdf/a only works for xhtml2, not xhtml

danbri: other things e.g. topicmaps could use it

danc: could just change schema documents and then harvest rdf out of them

tom: dc has an old spec for rdf in html using metatags. not sure how many people actually embed it. would need to check

ACTION: Tom Baker ask DC colleagues if many use rdf inside html

davidw: definitely a use out there - for transforming large volumes of web data

jeremy: what's the takeup of the note?

ben: feeling in the community that GRDDL is not 'offical'
... maybe all we need to do is endorse it but personally think people are waiting for rec track

gavin: if we could get all the rdf out of comment blocks then it would be very valuable

danc: feedback we've had is would we need to fix all the html? if so, that's a big problem

fabian: 90% of the documents we deal with are proprietory, so don;t embed it; for educational materials we deal with it would be good added value

ralph: chicken and egg problem - we're not seeig the demand for it because they don't know about it
... at best we can ask them to change once
... how fast do we think people whop want to put metadata in documents will move to xhtml2?
... dangerous to retrofit something to existing documents - the authors of the documents didn't necessrily agree to the new contract implied, especially if a random document

gavin: what would be the second change?

ralph: if we ask them to trty something experimental to see if it should become a rec, imples another change

jeremy: worth waiting another year for RDF/A, if only change once

pepper: if GRDDL worked with HTML not XHTML then it would really take off

danc: at the moment just uses xslt so won;t work
... not sure how it would work, need a parse for bad html

guus: not enough convincing usecases at the moment - would the TF provide those?

ACTION: Gavin find out from his community and contacts if they have usecases

danbri: was thinking what it would be like to wait for xhtml2 in order to use rdf in html: I'd choose GRDDL if have to choose one

jeremy: xhtml2 docuiments are immediately deployable because css can be read by existing browsers

Danbri: even the linking stuff? it would be nice if links between pages in xhtml2 still worked in existing browsers

guus: suggests TF waits for the input form those two actions and maybe goes to find more usecases itself

<danbri> jjc, that's v interesting; i'd like to see the detail if you've got a pointer

guus: we can use the final slot tomorrow to discuss it if we have any more information


Chris: I would like to propose to move two documents to Note
... first is "Representing Specified Values in OWL" -- aka "value partitions"

Guus: move to Note requires that the WD has been stable and any changes are minor

Chris: current WD was published on 3 Aug
... editorial changes are cleaning up terminilogy



<DanC_> (who else has read it?)

<libby> deborah

<libby> (sorry)

<DanC_> Representing Specified Values in OWL: "value partitions" and "value sets"

<DanC_> W3C Editors Draft 02 March 2005


<DanC_> (2 march... today's the 3rd)

<Natasha> I can barely hear what people are saying in the room

Natasha: the changes since 3 August are more than editorial

Chris: the content has not changed, just the way it is organized; just the structure

+Alan (earlier)

Alan: the only major change was 'feature' to 'quality'
... breaking things into bullets, revising the diagram
... put into the diagram the notion of disjoint by default

PROPOSED: to conclude work on Representing Specified Values in OWL: "value partitions" and "value sets" by publishing it as a Note, contingent on confirmation by Mike Uschold that changes since 3 Aug are editorial

<dlm> hand raising from dlm


Chris: moving on to Classes As Values
... changes in organization and wording
... Mike proposed some more editorial changes last night
... specifically, for the considerations sections, making each pattern consistent

<FabGandon> http://smi-web.stanford.edu/people/noy/ClassesAsValues/ClassesAsValues-2nd-WD.html

Mike: and rephrasing approach 4 to improve clarity

<Natasha> http://lists.w3.org/Archives/Public/public-swbp-wg/2005Mar/0029.html

DanC: has the WG shopped this around and gotten feedback that these are useful?

Alan: we got a flurry of feedback in September
... we've presented this in tutorials

ref earlier "value partitions" draft, the path to the current editor's draft is
-> http://www.w3.org/2001/sw/BestPractices/#Tasks SWBPD Home Page
--> http://www.w3.org/2001/sw/BestPractices/OEP/ OEP TF page "editor's draft")
---> http://www.cs.man.ac.uk/~rector/swbp/specified_values/specified-values-8.html
----> http://www.cs.man.ac.uk/%7Erector/swbp/specified_values/specified-values-8-1.html

DanC: what about names for patterns?

Natasha: I thought about it, but didn't finish

the path to the document currently under discussion is
-> http://www.w3.org/2001/sw/BestPractices/OEP/ OEP TF page
--> http://smi-web.stanford.edu/people/noy/ClassesAsValues/ClassesAsValues-2nd-WD.html Representing Classes As Property Values on the Semantic Web 2 March

David: there are open issues in the Status of this Document
... e.g. a promise to develop a dictionary of terms
... do you have a schedule for this?

Chris: yes, we are working on this
... we had hoped to have a glossary by this meeting

David: the highlighted terms are a to-do item

Chris: the highlighting and the to-do list will be removed
... when we publish the Glossary, we will make it consistent with the usage in this document

<DanC_> (ah... yes, having the TODO in there vs proposing to conclude work had me scratching my head too. moving the TODO list to the TF page would make sense to me)

David: what about the first bullet? ["identify several OWL DL compatible approaches..."]
... for a document that will live longer than this WG, I would prefer that the to-do list be removed

ACTION: Natasha remove the 'Open issues' from the Status of this Document

ACTION: Chris move Classes as Values and Value Partitions to w3.org

Mike: Alan highlighted pros and cons, that seems to be useful
... are people comfortable with using this approach in Classes as Values?

David: I found the pros & cons very useful

Guus: for Classes as Values I think there should be no opinion

Mike: not saying 'good' or 'bad' about the pattern overall

Guus: I feel strongly we should stay with a neutral approach
... this Note is about DL vs. non-DL
... it is dangerous to make subjective statements here

Mike: is saying "maintenance is costly" too subjective?

<DanC_> ah... "# There is a maintenance penalty"

David: saying "this is expensive to maintain" might require further review

Natasha: I would not want to make a judgement for everything
... some cases are obvious already
... I prefer a neutral approach

PROPOSE to accept http://smi-web.stanford.edu/people/noy/ClassesAsValues/ClassesAsValues-2nd-WD.html contingent on editorial changes to be proposed by Mike and accepted by Natasha

<dlm> no objection

(vote by show of hands)

Evan: this seems to be a convoluted process
... there seem to be substantial structural changes happening
... we're making a judgement about the nature of these changes

David: specifically, we just discussed changes to the value judgements in the document and decided not to make such changes

Alistair: abstain

RESOLVED to accept http://smi-web.stanford.edu/people/noy/ClassesAsValues/ClassesAsValues-2nd-WD.html contingent on editorial changes to be proposed by Mike and accepted by Natasha

<danbri-laptop> (re wordnet, i've not studied Aldo's new work)

Chris: the N-ary relations draft is still undergoing change
... this document will have content changes
... Ralph has the action to review this when it is ready
... also a new editor's draft on simple part-whole relations
... new draft co-edited by Alan and myself ready for comments by others

-> http://www.w3.org/2001/sw/BestPractices/OEP/SimplePartWhole/index.html Simple part-whole relations in OWL Ontologies 1 March

Guus: volunteer to review

Bill: volunteer to review

Chris: I hope soon after the OEP telecon 2 weeks from now that this will be ready for review

<DanC_> Last-Modified: Thu, 03 Mar 2005 16:05:53 GMT http://www.w3.org/2001/sw/BestPractices/OEP/SimplePartWhole/index.html

<DanC_> (though it bears the date 1 Mar 2005)

Elisa: will be spending time with Evan discussing units and measures note

Evan: I have a new task to work on this at NIST
... I've been looking at a lot of OWL ontologies now
... there are a lot of OWL ontologies for units, well fleshed-out
... first task is to produce a set of criteria for evaluation

Evan: the WG is not supposed to pick "winners", correct?

Guus: you could propose some minimal criteria for usefulness

Deb: GEO group has a starting point for units and measures in OWL also

Evan: based on ISO work?

Deb: not sure -- will ask

Evan: my intention is first to develop some evaluation criteria

ACTION: Evan and Elisa develop criteria for evaluating units and measures ontologies

<dlm> http://www.geongrid.org/workshops/geoont2005/ is the working group meeting I am at . they have starting points in owl for units and measures, numerics, scaling, and comparators

Guus: I see 3 types of things that could go into the note
... a generic schema for units and measures
... initial examples from Tom Gruber
... 2. actual units and measures themselves
... 3. patterns for using these; showing how to apply them

Guus: could concentrate on some at first

Elisa: there are hundreds of units we could consider
... so it would be helpful to narrow the scope

Chris: still waiting for Jerry Hobbs to join the WG [to work on time ontology]
... have pinged relevant AC Rep

<DanC_> (Chris, I'm kinda motivated to help with getting Hobbs to join the WG; I might have time to phone his AC rep)

further discussion scheduled for 12:00-1:00 EST tomorrow

David: what is the publication plan for the glossary? wikipedia?

Chris: not sure, will link somehow from WG pages; expect it to contain ~20 terms

Guus: maybe include as an appendix in future Notes


Guus: up until a few weeks ago we had little input
... now we have a lot of input; still processing it

Chris: I read one of the documents

<DanC_> from the agenda...

<DanC_> [13] WNET: Ontowordnet

<DanC_> http://lists.w3.org/Archives/Public/public-swbp-wg/2005Feb/0066.html

<DanC_> [14] WNET: WordNet data model:

<DanC_> http://lists.w3.org/Archives/Public/public-swbp-wg/2005Feb/0068.html

<DanC_> [15] WNET: ISLE lexical entries

<DanC_> http://lists.w3.org/Archives/Public/public-swbp-wg/2005Feb/0069.html

Chris: I read the 'mapping' document
... reconciling with good ontology practice
... recognized semantic issues with the toplevel of wordnet
... all good considerations from my point of view

Guus: good action list was developed at Bristol f2f
... would be valuable to merge this

DanBri: I am excited to see the ontologized approach moving along
... but I'm worried that we're lurching around; feels like independent academic research being reported to the WG
... how can we better work together to avoid 2-3 month gaps
... how does Brian's work fit with Aldo's?

Guus: Aldo's email suggests he is building on Brian's work

DanBri: I would like to see more of the discussion on the mailing list

<danbri-laptop> (I wonder whether a dedicated mailing list might help provide a place for dedicated wordnet/semweb collaboration...)

more on OEP, XML Schema Datatypes, Versioning

ACTION: Chris to ask Alan to take over the Qualifying Cardinality Restrictions Note from Guus

<Zakim> DanC_, you wanted to ask about previewing the XML Schema discussion, and noting some DAWG stuff

DanC: SPARQL has a lot of symbolic matching
... but if a variable binds to a number you can test, e.g. greaterThan
... a set of test cases is being written for SPARQL
... if SWBPD wants to get involved in this [ref. XML Schema datatypes], this would be a good time

JJC: I met for an hour this week with Don Chamberlin
... found no big disconnects
... the key issue appears to be at the semantic level of RDF datatype reasoning
... e.g. are "1.0"^^integer and "1.0"^^float equal? syntactically yes, real question is at the semantic level
... we're unlikely to reach an answer before SPARQL goes to Last Call
... I don't feel this open issue is a show-stopper

DanC: big design principle was to import from XQuery

<Zakim> DanC_, you wanted to give an example: does this query win or lose? ... AND "1/1"^^my:rational != "2/2"^^my:rational

JJC: in discussing with Don Chamberlin and the SPARQL editors, when we got to a hard question the answer in XQuery was "we structured the language so you can't ask that"

DanC: my example has more to do with open-world vs. closed-world reasoning

<DanC_> "1/1"^^my:rational != "2/2"^^my:rational

<DanC_> not("1/1"^^my:rational = "2/2"^^my:rational)

DanC: one design moved the not to the outside; inner returns False as my:rational was not recognized
... inner might instead return "don't know"

Stephen Harris: XQuery may use some variation on returning "don't know"

JJC: I also took an action to research how OWL-DL handles unknown datatypes
... I came away from yesterday's discussions [with SPARQL editors] without a lot of anxiety

DanC: ref. yesterday's plenary discussion about versioning, OWL has some things; might make this more visible
... could suggest to TAG to look at using OWL versioning for other things

<danbri-laptop> see http://www.w3.org/TR/owl-ref/#Header

Guus: OWL versioning was simple to do

DanC: but the solution is relatively unknown; might be worth an article

TomB: versioning seems to be in scope for Vocab Management
... would be nice to find a common mechanism that works for SKOS, FOAF, ...

DanC: TAG next meets in June

<danbri-laptop> (hmm OWL Versioning is mixxed up in the Full vs Lite/DL design... Annotation properties etc)

<danbri-laptop> [[

<danbri-laptop> <owl:AnnotationProperty rdf:about="&dc;creator">

<danbri-laptop> <rdfs:range rdf:resource="&xsd;string"/>

<danbri-laptop> </owl:AnnotationProperty>

<danbri-laptop> ]]

<danbri-laptop> etc

TomB: DCMI has a versioning model but we haven't yet figured out how to formally declare it

TomB: OWL would be one candidate way to express this

<DanC_> 7.4 Version information in OWL Reference

Chris: but you wouldn't be likely to use the built-in OWL versioning mechanism
... OWL version has no semantics; it is only annotation
... I expect [DCMI] wants to write a schema for versioning

DanC: I referred Henry Thompson to Jeff Heflin as the source of the OWL versioning design

Guus: that was correct

[adjourned to 0900 Friday morning]

XML Schema Datatypes

-> http://lists.w3.org/Archives/Member/w3c-xml-schema-wg/2005Jan/0027.html Re: XML Schema Datatypes and the Semantic Web [Dave Peterson 2005-01-31]

<MSM> http://www.w3.org/2001/sw/BestPractices/XSCH/xsch-sw/

--- xml schema problems ----

<RRS> XML Schema Datatypes in RDF and OWL

jjc: document has not changed much since the first version
... has been editorial changes, except for when to use what
... that is the only major change
... question: do we publish it?
... two comments: hayes' dawg review
... one small comment from peterson on a mistake i made
... the document was initially was for review the datatypes and to when are two values equal?
... the rdf and owl semantics do not specify that (for float and ints for example)
... xquery/xslt had a problem for the duration
... they were pulled out from the rdf/owl, but xpath2 solves that and the text refers to that
... section 5 is on the use of numeric types is for a different set of readers
... it gives some suggestion on when to choose what
... on the user dfeined datatypes the problem is: xml schema gives a mechanism to define his/her own datatypes
... owl/rdf design requires to identify datatypes with URI-s, xml schema do not necessarily do that
... the two will not work together
... we decided to postpone this problem
... there are ways to address this, the issue is more that it covers too many specs

<RalphS> definition of adultAge just prior to http://www.w3.org/2001/sw/BestPractices/XSCH/xsch-sw/#sec-daml-soln

jjc: daml+oil uses the name of a schema type uses the URI of the document plus # and the name to address an element
... that works, there are implementations, it is seriously non aligns with recs
... the problem is what is the frag id in general;, the architecture says that it is up to the document what the frag id is
... the daml+oil is nowhere close to what the xml or the xml schema solution might be.
... the second solution is to use xml schema component designators from the xml schema group, currently a stable working draft

msm: the binding vote should be close, close to be in last call

jjc: what we trying to do is a pretty basic use case to the xml schema component designator

<MSM> Current SCDs editors' draft: http://www.w3.org/XML/Group/xmlschema-current/SCD/scds.html

jjc: however, because of the generality of the solution, the frag id is pretty complex (based on xpointer)

(jjc shows the example in the document)

msm: the change we have made is to make the expression a bit simpler, it will look much more like an xpath expression

jjc: this is still more complicated than the daml+oil solution, and has difficulties when using n3 which uses qnames
... after the ':' n3 requires n3 names, which does not include '(' and others

<MSM> ...#xscd(/type(adultAge)) becomes ...#xscd(/type::adultAge) -- or, if the type adultAge is assigned to a namespace bound to prefix 'p', ...#xscd(/type::p:adultAge)

jjc: so it will lead to problems with deployed formats
... so it is ugly, but the generality is attractive
... there is an issue which is at the hear of pat's comment: according to xsd the component designators are for the simple type definitions

msm: no, the simple type definion is an abstraction

<PatrickS> Does it identify something that is a member of rdf:Datatype?

the abstractions are what a schema is made of

msm: the reason it took so long is that the theological work to say that we are pointing at the abstraction and not the xml
... i am not sure whether it was crucial but this is it
... specifically, the phrase "i.e. referring to the definition rather than to the type defined." is wrong

jjc: what is clear is that the way rdf/owl talks on datatypes means that the theological debate is probably unnecessary
... the simple case shouild work
... it does not seem that hard
... i have looked at the xml schema solution and the daml+oil

<MSM> The "simple type definition" is an abstraction (name, base type definition, facets); we call it a "definition" to distinguish it from the value and lexical space which follow logically from the base type and facets. (Distinction between intension / extension)

jjc: the rfc2316 says you take the url, you get a document with a mime type, and that tells you how to interpre the frag id
... the xml schema documents are xml documents (appl+xml) the mime type permits a bare name, there is a certain amount of deployed experience
... with bnare name, with a name after the '#' and that is an xml id, and that the frag id refers to the xml element

<RalphS> http://www.w3.org/2001/sw/BestPractices/XSCH/xsch-sw/#sec-id-attr

jjc: we could modify the xml schema to put an id on
... then the daml+oil solution is close to a solution. At least to me it does not seem to be so theologically unsound to do that

to use the the same uri to address the datatype described with that portion of xml

jjc: pat's makes a difference between the xml datatype and the note in the rdf semantics
... in my view the id solution is probably a good solution. the xml schema designators one is the general solution, but if you own the xml schema file, than using id seems to be better
... however, there is a theological debate around this
... that is the issue

guus: you propose is to say in our document is to stick an id into the xml schema if you own it

<RalphS> Jeremy's reply to Pat Hayes' comments

jjc: the document seeks opinions. My personal opinion is that the use of the id and the designator solution is optimal

timbl: rdf/xml says that its ID overrides the xml things
... schema would do the same thing

<RalphS> Pat Hayes' comments

timbl: so you cannot use the bare names you could not use it to refer to a chunk of xml
... would that be such a departure that it would horrify
... a schema could have a MIME type definition which defines that the fragids identify the type, not definition of it (like for rdf/xml)

msm: what could help the working group: what has thus far kept us to go is that registering a mime type is a mine field
... the procedure is now shorter, so that may be feasible, and I can take that back to the group
... but there is a concern: i can imagine wanting to talk about simple type definiton and the xml element used to declare it

if we make the ability to point depending to the mime type i loose the ability to refer to the xml element itself

msm: jjc's id solution means refers not to what we have but to something that is adjacent, because you know what it means
... in the strict sense it relies on a processor
... i would suggest that strictly speaking all of the bits in the scud (schema component designator): it is a problem that it is longer via xscd, but this tells you exactly via the xpointer mechanism what the exact id is
... it addresses the fact that it refers to an element adultAge.
... in any practical context, they add a prefix to distinguish a prefix to avoid name collision
... so I agree it is longer, but it is semantically simpler

Mike: i am all for to have semantically meaningful, can we handle that using a namespace

jjc: in rdf/xml you can use entities, and that works fine
... but in n3 does not have qname abbreviation
... if the last character is not a proper one

patrick: is there a way around that, updating n3?

timbl: ')' is used for punctuation, you could put an entirely URI there, you could talk to ROy to see how that framework works

patrick: it is really important not to loose focus on how people using datatypes can do that

<MSM> One possibly relevant fact: like XPath, SCDs will have short-forms, so "/type::p:adultAge" can be abbreviated to "/~p:adultAge" and "/attribute::p:adultAge" can be abbreviated to "/@p:adultAge"

patrick: the best practice using xml schemas should fit in a larger way of using datatypes in general
... you can use some other mechanism (java etc), we should also talk about abstractions and not only that particular processor

<Zakim> timbl, you wanted to suggest that regarding a schema as a higher-level langauge than an XML document is and to suggest that regarding a schema as a higher-level langauge than an

<danbri2> (I'm v intrigued by "The use of the # name also allows content negotation among many languages")

timbl: you can look at schema represents an infoset
... you can always go back to the source and make a link
... the schema defines types, it is not xml it is a schema language
... if they want to use it as xml, then it could be served as xml

(scribe has given up...)

msm: as long as there is a way to get 'back', you want to optimize and choose the opt. points wisely, so optimizing chooising the declaration rather than xml is o.k.
... but tehre are some unexploded mines:-(

<RalphS> MSM: typo in 2.4 -- should be base="xsd:integer"

msm: consider your example it points to another type (integer) for the restriction
... any processor would have a complete udnerstanding of adultAge
... the problem is when the base is not integer, but my:humanAge, for example
... if humanAge is declared in another schema document
... then depending how this is done, I may end up combining this with version 1 or version 2
... so the result is that the source declaration can end up with different interpreations

msm: strictly speaking the definition defines adultAge in context of the full schema
... we do not have version solution, so we do not have a good solution for this
... if the base type is an xsd: one or is in the same document, you do not ahve a problem

patricks: in the example you give, I would assert that is not a problem as long as you use different names
... you are talking about two different abstractions

ralphs: tim's notion to use mime types to explain the semantics of the document is fine, but maybe there may be a tag issue on whether the semantics are carried through the namespaces

<MSM> Patrick, yes, in principle. But consider situations like that of the HTML namespace. The abstraction 'p element' is (according to community practice) regarded as a single abstraction. But the legal contents are specified one way in the transitional definition, and a different way in the strict definition.

IvanH: do you want to use xml:id rather than id ?

<timbl> schema:id

<danbri2> (tim's right; it's a philosophical not theological discussion)

jjc: second issue is practially more difficult
... is the comparison of floats and decimals
... within xml schema there primitive types and the other simple types that are derived by restriction

<RalphS> http://www.w3.org/2001/sw/BestPractices/XSCH/xsch-sw/#sec-values

jjc: eg decimal is arbitrary long
... there are around 17 primitive tupes
... all the relevant specs are clear that when you derive from a type the underlying semantics does not change
... rdf/xml are agnostic on the issue whether 1.0 integer is the same as an 1.0 decimal
... xml schema is geared to a specific use case: schema processing
... in that case it is fine

msm: we are required to describe validation precisely
... the schema position is: yes these are quantitative values, 1.0 has obvious relationsships to 1
... they are not identical for schema purposes
... and applications may (eg, xpath2.0 operators) do more
... schema is a bit like an assembly language
... nothing prevents an application to define casting
... rdf can do that

jjc: solution #1: to do exactly what schema does, ie, they are different
solution #3: xml schema gives you a mathematical specification; use that

jjc: that would also be a very purist line, but may not be useful, it has surprises
... solution #2: xpath has solved the problem, they have defined 'eq', so use that
... although there are surprises because 'eq' is not transitive, for example
... it might be a show stopper
... at some level a choice has to be made
... we may be lucky and get a good feedback

guus: the schema people would prefer #1?

msm: none of the 3 solution would cause a problem for us

IvanH: DAWG has chosen the XPath solution
... it would be nice if SWBPD did not give totally different advice than DAWG

jjc: DAWG is doing something slightly different; they're specifying semantics of a query language, not of the underlying data

IvanH: they're defining a quality of the underlying resources

jjc: this may rule out the purist option

msm: strictly speaking what qt does is to define an operator called 'eq' and you may define it as you want
... so you could define a set of operators

<Zakim> danbri, you wanted to ask which xpath (1.0 or 2.0 datamodel) DAWG use, and whether that matters here?

danbri: the dawg guys are commited to xpath 1 or 2?

jjc: 2, they are using the operators of xpath2

guus: does not that influence your choice?

jjc: it will make it the xpath solution more attractive, but may not be the only one

guus: but it would be very strange if there are two different interpretations
... might be useful to talk to the dawg people on that

jjc: i had already some discussions, but not conclusive yet
... maybe we can add some indication of preference saying that a direction works better on sparql than some other
... but it would good to publish this soon

patricks: when you look at the options, it is important to get the best solution without breaking of owl reasoners
... if you choose only those that are safe for owl that would be good
... this chunk of useful equivalence is safe for owl reasoner
... i would encourage those that are involved in owl reasoners to comment

jjc: i think we could comment what we got after getting the comments of today

guus: I would like to publish now
... we could get general feedback

ACTION: Jeremy to incorporate the comments + pats' comments + peterson's comments

jjc: we could slightly change the intent saying that 'currently solution this or that is best'

guus: it might be clearer for feedback if editorial preference is listed. Evan sould review again

jc: realisticaly I would hope to get back end of next week

guus: would be nice to make a decision on publishing on the next telco

Mapping abstract model of WSDL

Bijan is given floor

Bijan: Speaking about WSDL => RDF mapping

Primary thing being mapped is abstract component model of wsdl

components have component properties that relate them to

sets of components or components

either a straight mapping where all the details of wsdl component model

are expressed in RDF/OWL

OR create a simpler model that glosses over some of the details of the wsdl model

but expresses the key concepts adequately

using the more faithful mapping wsdl-straight requires good blank node supports

in wsdl-straight property names tend to relate to plurals, whereas in wsdl-ont-nice a property name links

to a single component

Patrick: on cc/pp

do not underestimate impact of model on query

nice approach makes it easy to write queries

jjc: the difference is only mapping of forall contains

bijan: but then the english definition text differs from RDF model

timbl: keep it simple stupid, make mapping automatic,
... strip out all the sets

chrisw: why left hand side chosen?

i.e. the wsdl-striaght

there is also a Z notation for wsdl-straight

bijan: issues to do with faithful as in as close to transciption as possible
... possible way forward use nice model with straight model as informative appendix

RDF &Topic Map Interoperability

<pepper> slides for RDFTM: http://www.ontopia.net/work/survey-pres.html

<RalphS> (Valentina is a new member of the WG)

steve: estimated size of audience for survey is 50 people
... test cases not complete nor intended to be
... overview of previous proposals for RDF TM interop
... overview of evaluation criteria
... fidelity considered important

<RalphS> snapshot of Steve & Valentina's slides for RDFTM discussion

<danbri2> (hi-fi vs low-fi slide is useful; low-fi is a lowercase-r-reification of Tm structures, rather than RDF that carries the sense of the original TM)

bijan: conclusion semantic mapping more important
... survey vs. tutorial focus

mikeU: have a few statements and references

Steve: OK
... Coverage of OWL?
... used as it can help with translation

DavidW: was OWL considered in any of the other work?

SteveP: no OWL in surveyed work

DavidW: Then not covering OWL is OK
... if OWL can address open issues then say that
... want to ensure that "constraint languages" (Topic Maps) interoperate with OWL

<Zakim> danbri, you wanted to note that introducing OWL concepts/facilities to TM community (and v-versa) would be a useful contribution

Steve: Topic Maps Constraint Language is a current work item in ISO

DavidW: feel that "Introducing OWL to TM" would be out of scope
... for guidelines that is useful, but not the survey

<Zakim> RalphS, you wanted to say this isn't a tutorial

Ralph: note that OWL might have helped somewhere, (as said above)

SteveP: OK to mention commercial implementations?

<RalphS> put a sentence in the appropriate places for each approach where it could have been improved by using something from OWL

RalphS: matter of degree. OK to say at least once that there are implementation and cite them,
... but repeated reference may be overboard

DavidW: for commercial implementations "see this reference" as opposed to mentioning it inline,
... esp. given editor is from the company mentioned

discussion of objectivity given mentioning of commerical implementations

guus: should be very clear that its an opinion section, separate from "objective" section

steve: will look at specific sections that seemed subjective and discuss

<RalphS> Natasha's mail

steve: will ask Natasha for specific sections

dan: important to be fair,

Danbri: the document could explicitly solicit pointers to other implementations via the mailing list, which would increase openness

steve: would like to consider input that has been published

Steve: I think we know about everything that's out there

steve: anyone not convinced for semantic mapping (one conclusion)

davidW: i buy the argument, but unclear on editors position after survey
... how far could semantic mapping go in addressing problems

steve: semantic mapping is the only way to go, but don't know if 100% complete
... not sure a top priority

danbri: one style uses same namespace URI, the other shadows a similar one

steve: reusing vocabs and therefore URIs seemed preferable

<Zakim> danbri, you wanted to distinguish 2 kinds of semantic mapping

guus: on objectivity treat it in a mechanical fashion - this is our job, this is our approach - helps remove subjectivity
... can't imagine a solution approach in which OWL would not be helpful

Steve: agree

review of PFPS comments from 2001

<RalphS> RE: On the integration of Topic Maps and RDF [Peter Patel-Schneider 2001-08-01]

steve: shoudl two test cases have indentical information content
... think so, danB will write some

<RalphS> RE: On the integration of Topic Maps and RDF [Patel-Schneider 2001-08-21]

mikeU: may be forced to constrain one side based on the expressiveness of the other

guus: danger of having contrived test cases

<RalphS> RE: On the integration of Topic Maps and RDF [Martin Lacher 2001-08-23]

steve: test cases in survey intended to be informative regarding naturalness
... danb should be able to do a good job about expressing knowledge in RDF "naturally"

mikeU: could have middle ground by having same core example, and then growing the example in multiple directions depending on capabilities

steve: want to keep examples short for survey

ralph: like shortness

steve: started to develop some guidelines test cases

<RalphS> brief examples are fine to illustrate each of the approaches. for the final summary and choice of preferred approach, then it would be good to have a more complete example

danbri: even small examples can explore huge problems (from experience)
... some difficulties arise when using datatypes or URIs

ralph: some of the approaches are so obviously flawed that it doesn't make sense to beat on them with test cases that include data typing, etc.

steve: consider moving test case results to separate document?
... prefer leaving them in

ralph, guus: up to you as editor

steve: are issues identified "requirements", if not what are they and where should they be documented?

<RalphS> for the purposes of this survey, if the audience is really 50 people, moving test results to a separate document feels like editorial busy-work to me

mikeU: nice to keep requirements separate
... but OK to start with looser ones and tighten up close to finishing

steve: starting to understand what issues will lead to requirements
... for example range/domain constraints not in topic maps, in RDF, how to we address those
... prefer to wait a bit and let the requirements arise as we work

mikeU: themes in considerations that are already there in document, could serve to capture that explicitly and introduce them in the beginning
... proposed a similar thing in classes as values note
... minor presentation thing to make easier to read

steve: OK
... importance of "naturalness" and "fidelity" (are they the same thing)?
... naturalness is "faithfullness" to the paradigm
... fidelity the correctness of the translation

mikeU: agree with idea of naturalness, but not made explicit in document. The word doesn't capture what you mean
... deeper problem is readability, when you use "semantic hacks" they don't translate well
... naturalness is not that important

guus: also much more subjective
... how do you measure that

steve: can you measure readability?
... agree w/ Mike, added a paragraph that discusses interoperability as it is impacted by "naturalness"
... e.g. using semantic hacks

danb: hard to measure a lot of these things, and good idea to focus on the terminology here

<Zakim> danbri, you wanted to prefer "natural" (or "faithful") over "fidelity", since latter appeals more strongly to concept of truth

steve: ok
... acceptable to require mapping information?
... key problem is that TM and RDF have different levels of semantics
... any triple could map to 6 different things in a TM
... can't know which unless you "understand" the predicate
... can get some information from the nodes
... acceptable to require mapping information? some believe can't be done w/o that
... sometimes can get information from a RDFS or OWL ontology
... do we need to be generic, ie applies to any rdf model, or require some semantics (ie in RDFS or OWL)

ralph: "required" is a difficult SW thing
... can discussed where this information might be if present,
... ie in the namespace document
... but requiring it doesn't seem so good

steve: take foaf:name - w/o semantic information would map wrong using a default mapping
... but if foaf:name was a subproperty of rdfs:label, would work better

jjc: two issues I see
... wrong means (to me) contradictory, not "not the best"
... annotation on a schema may be third party, keep in mind open world about where annotations come from

danb: good point, grddl does this
... we are deciding who to make work for, vocab owners, app builders,
... prefer to focus on smaller group
... happy to require mappings, making sure they are consistent

<RalphS> when work has to be done, it's better to require it of the vocabulary owners not the (more abundant) vocabulary users

steve: would like to finish survey and move to guidelines quickly

guus: timeline for survey
... get consensus by next telecon (Mar 24)
... think its important to get TM feedback

<RalphS> release a new editor's draft a week before the telecon -- i.e. 17 March

steve: mar 17 is doable

guus: once that is in WD, OK to work on guidelines
... parallel is OK, too, can start today
... make finishing survey a top priority

steve: OK. That's our goal
... need evidence that he approves of this

valentina: will communicate this back to fabio

ralph: needs to be in mail archive

jjc: would rather have TF discussions on list, even in italian
... responds in italian w/o using colorful italian idioms

guus: important to make clear that process is open

steve: OK - partially due to under-familiarity with process

danb: skos has its own mailing list but archived at w3c

steve: thanks. Our goal to have guidelines ready for "extreme markup" conference in Aug

<danbri2> (we didn't discuss relation to http://www.w3.org/2004/02/skos/core/spec/#subjectIndicator ... maybe at lunch?)

<Zakim> danbri, you wanted to speak in favour of putting work onto schema authors over app developers and content consumers

<aliman> .. re subject indicators also see http://www.w3.org/2004/02/skos/core/guide/2005-02-15#secpsi

<Valentina> hai seguito la discussione? Potevi chiamare... :)ù

<Valentina> cioe?

TAG Issues.


(jjc talking)

discussion of http range

everybody agrees that when you do a GET on an http URI you get a representation of a resource

jjc: dc:creator is a URI without a #

<DavidW> SUBTOPIC: HTTP range (# vs. /)

jjc: one school of thought says that, because slash URI is gettable, it necessarily ... scribe lost

timbl: http scheme is a scheme of documents ...

jjc: can slash URIs be used for abstract things ?
... for this group the issue is important because dc & foaf use slash, but if http range goes with hash these things are broken

timbl: http slash uris necessarily denote documents (information resource)

jjc: interested in published subjects
... http range decision breaks pubsub

aliman: says no it doesn't

<DavidW> The TAG refers to this issues as "HTTP Range 14": http://www.w3.org/2001/tag/issues.html#httpRange-14

pepper: we have one class of things: resources (RDF speak)
... other class of things that have location (addresses)

timbl: information objects have information content

pepper: information resources necessarily have an address

timbl: but does the bible have an address?
... info resource is not necessarily addressable

pepper: direct and indirect identification of subjects
... but web has no mechanism for disctinction

timbl: no, indirect identification we can do

example of the man who's name is fred

cf. use the URI to directly denote fred

davidw: if swbpwg has consensus on this issue, then Ttim needs to come in a defend his position
... if Tim not available, who can proxy?
... actually probably me (davidw) ...
... when we looked at this in tucana we used hash uris
... but practical issue how to deal with mature vocabs that wouuld be invalidated

jjc: also problem of large vocabs - large download problem

timbl: no reason to break up that document
... suggest you use sparql

or here's the algorithm to get a bit

jjc: so we could break up wordnet to make retrieval doable ...

but this seems to misrepresent knowledge

timbl: no keep the same namespace e.g. cyc can be broken up into chunks

david: way to subdivide the namespace?
... jjc said if you want to further divide a namespace you use a slash

jjc: if we define wordnet namespace with a slash ... (lsot)

timbl: in webarch uris identify the files ...

in semweb architecture uris identify concepts

phil: we're saying there's no space for duplicity ....

e.g. MIME type, interpretation depends on context

timbl: URI identifies one thing only

pepper: shows slides how single URIref can be used to identify two different things ...
information resource is by definition network addressable ...
therefore you can use the network adress as the identifier
but can also use the same URI as subject indicator ...
whatever you mandate people will use both hash and slash

timbl: wants to define a transition strategy to move foaf & dc to use a hash
... even if it involves building those two URIs into every single RDF parser

danrbi: if WG writes note, would tag review it?

timbl: one of tag issues is written up as an argument tree ...
so there should be a paragraph number for your position
so tell me where you got to.

davidw: tim has a string opinioin which he has documented and which he has persuaded others in TAG ...
... all has been dealt with ad nauseum ...
... therefore we should read the existing decision tree and read all other arguments before we re-invent the argument wheel ...
... so before we take a position we should read everything !!!!!

tomb: tim's proposal would invalidate so many things for DCMI ...
lots of guidance documentation would have to be rewritten
therefore tomb says timbl's strategy would invalidate lots of DCMI

<danbri-laptop> anyone got the url for timbl's position tree diagram?

jjc:anyone else feel they are up on the issue?

Alistair:what if http mandate is not enforceable?

david: we should review the decision tree

<danbri-laptop> (background: "What do HTTP URIs Identify?" )

Alistair I've recreated all these points over the past 6 months
... I'm worried about the social process of getting everyone to adopt a new solution
... each of the 3 philosophies feels consistent to me
... 1. Tim's
... 2. published subjects
... 3. "you can identify anything with http: but if it's not an information resource you should do a redirect'

<pepper> (tm background: http://www.ontopia.net/topicmaps/materials/identitycrisis.html)

david: philosophical issue: should w3c follow or lead?

<danbri-laptop> (SWBP might take the position that dc:title and foaf:Person terms _are_ information resources)

EricP: subject identifiers are pretty close to inverse functional properties in OWL

patrick: reiterate that all of the options are coherent, selfconsitent models ...
are all consistnet with current webarch also ...
question is not whether they are reasonable ...
but whether if we choose one over the other what will we break and what will we improve
bottom line is that industry has already decided - the semweb poster examples all use slash
and tim's approach to go to hash only is just far too expensive
so issue should be finally decided in favour of slash

pepper: it's a mess, too late to fix it, pragmatic issue, cannot force people to do something else ...
what happened to fragments? how to identify a fragment of a document? this is what the hash was designed for.

ralph: sympathise with tim's pain ...
conversation seven years ago, tried to persuade tim to tell us what he thought we should do ...
answer led me to encourage model & syntax WG to use whatever they wanted to use ...
but our understanding of these architectures evolves over time ...
tim has articulated a new position since seven years ago ...
things have tightened up since then ...
tag has not yet reached consensus because has representatives for lots of communities ...
perhaps strategy fwd for us is to recognise (1) there are existing applications that have made choices, and it would be unwise to try to get them to change ...
(2) but can say : from some time fwd the best practise is foo
but still don't force people to change

<danbri-laptop> note: Adobe XMP use /, see http://mitglied.lycos.de/virtuoso5/xmp/xmp.txt example

<danbri-laptop> xmlns:xap='http://ns.adobe.com/xap/1.0/" xmlns:pdf='http://ns.adobe.com/pdf/1.3/' ...

<jjc> and

<jjc> <pdf:Producer>Acrobat Distiller 6.0.1 (Windows)</pdf:Producer>

pepper: but what about fragments?

ralph: not our problem

danbri: would we begin best practise or would we declare best practise

ralph: our responsibility to look very carefully at tag record

david: one option to start a TF?

ralph: joint TF?

phil: these are observations: ...
the point about fragments is very relevant ...
because you can conceptualise a fragent of a document to be a concept ...
(can see merit in tim's point) ...
second point relates to lead or follow ...
we are a community of leaders ... cf. community of users (follows) ...
we should look at the community of leaders ... examine their position.

david: there are times to lead and times to follow

<Zakim> aliman_scribe, you wanted to say in SKOS Core to say about skos:subjectIndicator

<Zakim> danbri-laptop, you wanted to propose exploring position that vocabulary terms are "information resources" in just the sense of timbl's http://www.w3.org/DesignIssues/HTTP-URI.html

jjc: tag is divided

danbri: two things to say: ...
1. if we get this wrong we have a deployment disaster on our hands ...
lots of stuff has been written ...
cf. experience of dc namespace change and how long it took for chnage to propagate ..
dc dcterms foaf foaf-extension adobe xmp creative commons all use slash
2/3 - 3/4 of deployed semweb already uses slash

<RalphS> DanBri: XMP, RSS, ...

if we say: "change" without a compelling story we look stupid

david: and we slow down deploymnent

danbri: foaf files are interesting because they link to lots of other vocabs ...
could possibly get foaf users to change foaf, but then all the others too ... ?

jjc: but there is no compelling story

danrbi: we need to appreciate the scale of the problem, several people fulltime for at least a year ...
if we get it wrong we hurt semweb ...
alot of foaf stuff comes from perl scripts ...
but adobe have shipped applications - cost of change huge for them ...

phil: if we get it wrong and we get it late we hurt semweb

danbri: compromise position: fresh start for new namespaces
... also this topic is discussed in other fora ...
we are only concerned with uris for terms in semweb vocabs ...
could say that dc title and foaf terms are possibly information resources ...

david: summarise ... it's too late to lead ...
live with what we have, if we force into lead then none would follow .

tom: remember when we established namespace policy ...
wanted to establish it without the 1.1 not in the URI string ...
but without that comment was we would compromise the integrity of dc, evidence for instabliity etc.
so trying to explain a change of this magnitude would be *extremely* difficult ...
it's a philosophical argument, would take alot of resources ...
to make a change, but it would be a waste of resources, we have more important things to do ....

patrick: 1: (melodrama) if we say: though shalt use hash, this would require significant corporate support ...
but it would receive significant corporate obstruction from not just nokia ...
its about efficient access esp for low bandwidth devices ...
2: the way you present you document makes the difference ...
WG should do a 'not bad practise note' ...
say: look, there are proven, well established practises in semweb, here are usecases and benefits for each solution, because question is what is best when? (not either or) Nokia's position is that http URI can be used to identify anything .. and should not be any redirection ... so need eficient representation mecahnsim.

ralph: danbri said deployment disaster ...but we should distinguish between those changes that would require existing deployment to change ... and those that don't but the other side is ... in what ways would existing deployemtn break if we recommend a new model ...? exiting apps would continue to function ...question is architectural truth and beauty vs. practical engineering. timbl is about truth & beauty & model theoretic consistency but its the engineers that build the thing ...

<danbri-laptop> (my point on 'looking stupid' is not w3c group losing face, but the knock-on effect for the larger community around us who have championed the use of RDF these last 7+ years; they will feel betrayed, i fear...)

patrick: don;t think generalised view is not just truth & beauty, its about a particular truth & beautyeach one is consistent in itself and ... this WG SHOULD NOT USE 'SHOULD'

pepper: another problem with hash: server-side processing cannot be done with hash ... recent project defined 75000 terms ....went with slash because you can do server side processing, so cannot resolve these things

Ralph: tim admits that this is a bug

<jjc> Proposed question for straw poll: can an http URI (without a hash) identify an RDF property?

<RalphS> like metadata vs data -- what is an insignificant difference to one community might be another community's important data

<danbri-laptop> +1 to pepper's point; same happens w/ wordnet-as-classes

phil: observation: lots of past good work, but here too much discussion about the past and not enough discussion about the future, so beware!!!

ralph: diff communities have different priorities

david: consider jjc's question ...

<DavidW> "MAY" to be interpreted as described in [RFC 2119].

jjc: rephrase as "Should the WG say an http URI (without a hash) MAY identify an RDF property in a conformant way"?

<danbri-laptop> "May ... identify in terms of http://www.faqs.org/rfcs/rfc3986.html (Berners-Lee et al) "

patrick: is it proper, given current standards & webarch, to use a slash uri to identify an RDF prop


(including observers)

yes: 11

no: 0

what does conformance mean?

david: wg reconvenes, session on tag issue is adjourned

David: several points were made about server-side processing and the impact of certain URI usage

<RalphS> (TimBl left a while ago)

<danbri-laptop> draft idea: The WG believes that the practice of identifying RDF/OWL terms and vocabularies with non-# HTTP-URIs is consistent with http://www.w3.org/TR/2004/REC-webarch-20041215/ and RFC 3986. It notes that such practice is very widespread, but that there remains some uncertainty in the W3C community on this topic and that this uncertainty is having a damaging effect on SW deployment efforts.

ACTION: Jeremy draft text for statement to TAG reflecting the opinion of the httpRange-14 breakout discussion

<danbri-laptop> ref also http-range-14

<timbl> Jeremy, the note will presumably describe an alternative architecture, and how it affects existing applications?


Guus: Have a specific problem to be addressed: session at W3Conf -> need presentation showing two applications.

Fabien: I provide presentation with recorded demos of a number of applications present in the blog.

Steven Harris: a list of projects is available at http://www.aktor.org/technologies

Guus: should we go for a DOAP format?

Libby: criteria are really needed and should be clear (also opinion of Eric Miller) example : open source only - may be too restrictive. Should we include resources for developers only? Resources for promotional exercise? Just examples in general? Is it possible to constrain the audience? Guus you are the one to use it a lot?

Guus: Distinction like Company uses semantic web for internal systems. (close environment) versus used in open web environment.

Ivan: Fujitsu report is internal for instance?

Guus: Aerospace industry example ... no longer existing.

Bill: the internal vs. external distinction may not be relevant because an internal application may be affecting hundreds of peoples behind the firewall of the company.

Steven: a distinction could be "do you control the data?"

Libby: the real problem is to define the criteria for including/excluding someone in/from the repository

Gavin: it seems we are still trying to find criteria to narrow the scope. We must make the difference with finding criteria/attribute to view/navigate/sort database.

Libby: provide the list in OWL/RDF and readable format => list is unreadable => facetted browser would be much better. Not even sure we have to scale it down really.

Guus: application domain is another criterion.

Bill: yes it's natural / sensible ... for instances application is doing data mining / etc.

Guus: yes but also application domains e.g. medical domain / product selling / ...

Bill: protect privacy, etc.

Fabien: I use the application domain to answer questions for instance for the W3C communication (I am meeting wit someone in Bioinformatics, what semantic web applications do you have in this domain?)

Gavin: aren't there other directories we could learn from, the way they categorize, from their schemas, etc.

Ivan: there are some e.g. Semantic Web Board, etc. afraid of duplicating
... Dangerous path to count too much on W3C endorsement, and what happens when the TF stops?

Guus: ok let's stop the TF :-)

Ivan: who will maintain?

Libby: use DOAP and leave it to the users to maintain it.

Gavin: why do we need this repository?

Guus: a list of application is the most frequently asked question by people!

Ivan: is it possible with DOAP to build a web site that is maintained via some community effort?

Steven: that's what we do in our project (AKT project) list of URL of descriptions scanned every night.

Bill: we want no maintenance.

Libby: we may be too ambitious. Small descriptions (a pointer and a sentence) only with a pointer and people can use it to harvest.

Ivan: but somebody has to do it? What will happen in three years?

Fabien: two different problems: get a list and find a way to maintain it after the end of the TF/WG.

Libby: may be we won't need it in three years. :-)

Gavin: for the description how we get it?

Fabien: just from the form to generate your DOAP file.

Gavin: simple interface to accept the submission that could be used by someone else in three years.

Ivan: if this something that could lead to a significant collection not only for us geeks?
People are not interested in geeky stuff like FOAF, they are interested in the Photoshop example.

Gavin: distinction between project and products. Our product line use XMP can I put everything inside?

Ivan: as an admin I would say please put only one item.

Gavin: pointer to a technology vs. project vs. product

Guus: is there a particular type of selection to show the added value?

Ivan: still in the phase where we have to convince people that there are a lot of applications out there.

Libby: may be we should just focus on getting a number of them in the list. First priority.
Gavin & Ivan: only members should be able to put commercial products.

Libby: if require the RDF description that may slow-down the flow of descriptions.
Also recall that the criteria have to be precise.

Gavin: true if we could come up with the right criteria in the first place, but that won't happen. So it is not that important.

Libby: RDF and OWL applications only?

Ivan: for the time being only RDF and OWL.

Guus: I agree.

Bill: concerning the classification, right now we have not so many applications, so we may be trying to classify in the vacuum.

Fabien: an extensible flat list where users can add missing domains would be ok, if it grows then we can reorganize it latter (topic ontology :-))

Libby: summarizing = we stay with web log + we try to set up some tools for task force administration tasks (accept a description) + provide support for DOAP + maintain a simple list of DOAP files in the blog.

Andreas: will the list be available in RDF.

Ivan & Libby: Yes

Libby: the RSS blog gets picked-up by Planet RDF.

Andreas, Ivan and Stephen:
RSS is not really an RDF application since there are syntaxes in just XML (not in RDF)

Guus: but it cannot be ignored as an application that uses RDF.

Bill: cannot submit just an ontology; the application submitted must actually do something.

Ivan: with Mozilla it's your private data in RDF.

Pepper:do TM apps qualify?

Ralph: I propose that any Topic Maps application that supports our translation mechanism be accepted to the ADTF index

<danbri-laptop> ...UML, LDAP, ...

<DavidW> <rathole>HERE</rathole>

<Zakim> danbri-laptop, you wanted to sympathise with the TM case, but note that plain XML, MathML, KIF, Prolog, all have a case for this

<danbri-laptop> (I might've also noted that RDF itself could grow and mature... an RDF 2 might have strong KIF/CommonLogic, CG and TM influences... but RDF remains the architectural focus)

Chris: believe the ADTF page should be limited to apps that work with RDF & OWL

<Zakim> FabGandon, you wanted to talk about CG

<jjc> how about restricting to applications have web pages that validate as (X)HTML?

<jjc> or provide correct use of language tags?

<jjc> (jjc asides above)

Fabien: if we include Topic Maps apps in this list I will not be able to maintain my position to the Conceptual Graph community that they must support RDF & OWL

David: what is "free" -- as in "speech" or as in "beer" ?

Libby: "free" meant "not costing money"

Phil: feel there is space for a soft line
... we could talk about the "Web of meaning"
... in WOM there is space for other stuff

Chris: this is exactly what I would like to avoid

<danbri-laptop> some W3C Activity statement excerpts: "The goal of the Semantic Web initiative is as broad as that of the Web: to create a universal medium for the exchange of data. It is envisaged to smoothly interconnect personal information management, enterprise application integration, and the global sharing of commercial, scientific and cultural data" ... "The principal technologies of the Semantic Web fit into a set of layered specifications. The current componen

<danbri-laptop> ts are the Resource Description Framework (RDF) Core Model, the RDF Schema language and the Web Ontology language (OWL). " "The Topic Map (XTM) and UML communities have been finding increasing synergy with the RDF family of technologies." -- http://www.w3.org/2001/sw/Activity

<jjc> http://lists.w3.org/Archives/Public/public-swbp-wg/2005Mar/0054.html (draft MSG)

Chris: a lot of people in this WG are working on other technologies that are not specifically RDF&OWL but that are related; e.g. KIF
... you expand the charter of the applications page significantly if you force them to deal with these issues

Mike: could avoid the issue by titling the page "RDF And OWL Applications"
... as a theoretical point of view, there is no reasonable grounds for saying something that is not RDF & OWL is not the Semantic Web

Gavin: will there be a discussion 6 months from now on what goes on a "Semantic Web Applications" page?
... have we done anyone any service by taking off the label "Semantic Web"?

Steve: Can the line be drawn at semantic technologies based on XML and URIs? I joined the WG specifically because the W3C Activity Lead wanted my community to be part of the Semantic Web

Mike: it's not practical to say we can draw the line to include Topic Maps [and exclude others]
... RDFTM is a Task Force in this WG so therefore this is being considered
... we can choose to catalog just RDF/OWL applications as a way of bounding our work

Gavin: this would solve the immediate problem but not solve the larger problem of what the Semantic Web is and isn't

DanBri: this is not a static situation
... EricM is a very inclusive fellow; he and others go around trying to connect communities
... considerations from Topic Maps and others influence what the Semantic Web is
... the context we are chartered in is RDF, RDF Schema, and OWL
... the future of the Semantic Web will be more "Topic Mappy"
... you're here to do the mapping

Steve: does every Topic Map application have to support RDF directly?

DanBri: once we have a WD out we can say the entire universe of Topic Map activity joins RDF

Gavin: when talking about mapping into Semantic Web, what does this mean?

DanBri: being able to run SPARQL queries against data that was published as a Topic Map

Gavin: does this make Topic Map part of the Semantic Web or the transform part of the Semantic Web?
... if GRDDL is adopted, does HTML become part of the Semantic Web ?
... I'm trying to understand whether it is the transformation that becomes part of the Semantic Web or the technology that was formerly outside?

Guus: I prefer to stick with RDF/RDFS/OWL defining the scope of ADTF

<danbri-laptop> aside: the phrase 'lowercase semantic web' is being used by some for XSLT-able xhtml markup that carries semantics, eg see http://www.tantek.com/presentations/2004etech/realworldsemanticspres.html

<danbri-laptop> (Tantek was here yesterday)

Guus: PROPOSE that ADTF registry be limited to RDF/RDFS/OWL applications
... and we discuss on a future telecon what is
... and we discuss on a future telecon what is "part of" the Semantic Web

<dom> I just wanted to note that the log of applications is also limited to "free" applications

<dom> ... which isn't saying that non-free applications are not part of the SW

Chris: my goal was to scope the work of the ADTF

<dom> ... that's only called scoping a problem, AFAICT

Message to TAG regarding httpRange-14

Possible DRAFT msg to TAG [Jeremy]

<jjc> s /primary concerns are/primary concern is/ in DRAFT msg

<Zakim> pepper, you wanted to add the rape of the concept of "fragment" to the list of other important concerns

JJC: I would object to adding fragments to this as it gets more philosphical

Steve: inability to identify document fragments is an issue

Dave: it's less important for us to address Web Architecture issues than to address Semantic Web issues

<dom> [I don't see why this makes it impossible to identify document fragments]

Dave: I don't want to cloud the issue with things that are not crystal clear

DanBri: the more we can do to narrow down to a closeable part of the problem space, the better
... we don't care whether cars and airplanes can be identified with http: URIs; this is about RDF properties

Alistair: the tone of this message will make people on the other side of the debate dig in their heels
... the issue needs to transcend the opposition

Guus: this is a very factual message

Chris: suggest dropping "failure to resolve" and just leave "This issue is impactin..."

<jjc> Suggestion to drop final para

DanBri: this debate should not be allowed to go on for another 2-3 years

Guus: we may not be able to reach consensus on this today
... may need to postpone to a future telecon

BenA: my impression is that any argument we give that is based on "it would be hard to redeploy" would not help -- Tim would not be receptive
... technical arguments would make a stronger point

Guus: my assignment is to chair a deployment group
... if deployment issues don't count, why are we here?

David: this message is going to the TAG, not specifically to TimBL

Patrick: this is a request to the TAG for closure, not consensus
... the TAG can close on the issue with dissent (if necessary)

location & time of next f2f

Mike: prefer Seattle area or Galway (concurrent w/Sem Web conf)

Jeremy: prefer to have it outside the US
... due to one member not being able to attend this meeting due to visa issues

Steve: Montreal in August is one possibility

<danbri-laptop> +1 w/ jjc's concern

Steve: w/Extreme Markup conf

Gavin: Ottawa is another option

<libby> +1 on extreeeeme

Gavin: my office is in Ottawa

Mike: I can look into hosting in Vancouver

<danbri-laptop> (I abstain re whether i can attend any non-Europe travel: funding uncertainties... I can pay my own way to Galway happily enough I think)

Andreas: DERI would likely be willing to host in Galway

<danbri-laptop> see http://iswc2005.semanticweb.org/

<danbri-laptop> (for bar discussion: if http://iswc2005.semanticweb.org uses Javascript for hyperlinks, is it a Semantic website?)

<dom> (I would argue that they don't use enough declarative sematics to qualify as such)

Mike: is November too late considering the WG charter ends 31 Jan ?

Guus: October would be nice

<jjc> I will find it easier in November,

Guus: I'll do Web poll on Vancouver & Galway

---- closing summaries ---

Guus: there were 17 documents on the reading list for this f2f
... we should all be glad with the progress over the last 4 months
... we should look forward to closing some of our Task Forces; this would be a positive
... glad to see nice collaboration on UML and Topic Maps

PROPOSE to cancel 10 March telecon, next telecon 24 March

scribe:make 24 March a 2-hour telecon?


Summary of Action Items

[NEW] ACTION: Phil to update SETF charter in light of new focus for SETF draft note
[NEW] ACTION: TomB to post URL to his VM TF slides
[NEW] ACTION: Ralph to inform the W3C Communications Team that we intend to cite http://www.w3.org/Consortium/Persistence as "W3C URI Persistence Policy"
[NEW] ACTION: Alistair to change the wording of the link to "the persistence policy at URL http:"
[NEW] ACTION: Alistair to think about machine-readable change policies
[NEW] ACTION: Alistair to change the links to examples
[NEW] ACTION: DanBri help write an rdf schema for the additional xhtml2 namespace elements
[NEW] ACTION: BenA set a time for the RDF-in-XHTML telecons
[NEW] ACTION: the rdf in html tf to discuss whether GRDDL needs to work on XHTML2 documents
[NEW] ACTION: Tom Baker ask DC colleagues if many use rdf inside html
[NEW] ACTION: Gavin find out from his community and contacts if they have usecases
[NEW] ACTION: Natasha remove the 'Open issues' from the Status of this Document
[NEW] ACTION: Chris move Classes as Values and Value Partitions to w3.org
[NEW] ACTION: Evan and Elisa develop criteria for evaluating units and measures ontologies
[NEW] ACTION: Chris to ask Alan to take over the Qualifying Cardinality Restrictions Note from Guus
[NEW] ACTION: Jeremy to incorporate the comments + pats' comments + peterson's comments
[NEW] ACTION: Jeremy draft text for statement to TAG reflecting the opinion of the httpRange-14 breakout discussion
[End of minutes]

Ralph R. Swick <swick>
Minutes formatted by David Booth's scribe.perl version 1.117 (CVS log)
$Revision: 1.14 $ of $Date: 2005/03/23 19:38:14 $