SWD Boston Face-to-face, day 2

23 Jan 2007

See also: IRC log, day one minutes


Jonathan Rees (Science Commons), Ivan Herman, Ralph Swick, Michael Hausenblas, Antoine Isaac, Guus Schreiber, Steven Pemberton (HTML WG), Tim Berners-Lee (W3C), Alistair Miles, Diego Berrueta, Tom Baker, Ben Adida, Mark Birbeck (HTML WG), Kjetil Kjernsmo (Opera), Fabien Gandon, Stephen Williams (HPTI), Elisa Kendall, Alan Ruttenberg, Daniel Rubin
Antoine, Fabien, Alistair, Kjetil



RDF Use Cases - http://www.w3.org/2006/07/SWD/RDFa/scenarios/

Ben: we are continung work with RDFa syntax, primer and UC doc
... compiling rdfa test cases
... keep in mind that html wg is still under review

<Steven> Review ended last friday

Ben: we expect that the rdfa syntax can be adapted to every html syntax
... adapting test cases etc

Ben: Overall schedule: push primer further with help of WG, and use case
...syntax and html module within 6-8 weeks

Ivan: if it is W3C Recommendation track, end of february would be for last call?

Ben: no. That was not the idea for now
... last call means technical issues are addressed
... we still expect reactions and comments on various aspect

Guus: two issues
... pushing things further with swd wg
... and html aspect
... propose to bundle
... can this wg publish a rec on a module for html?

Ralph: our charter allows for that
... the question is still open

Steven: dropping html wg: existing one or proposed one?
... if existing one, no problem with publishing a rec

Ivan: there are precedents of modules published by other activities

Mark: it's not a modification of html
... it's a module which uses XHTML 1.1 M12N techniques

Ben: bundle idea seems good idea to me

TimBL: importment is deployment strategy
... pushing RDF into attributes. Current browsers do nothing with attributes

Ben: yes

TimBL: other issue are HTML tidy and validation

Ben: content management tools need publishing and validation, yes
... I don't want to discuss the syntax
... I want to discuss rec or not

Guus: my feeling: discussion on UC is different if we go for rec track

Ralph: short term question is readiness to publish new version of doc
... how much is needed depending on our choosing note or rec

Guus: would the UC doc content need to be different?
... postpone the discussion, ack that doc might end in rec track

<benadida> use case doc: http://www.w3.org/2006/07/SWD/RDFa/scenarios/

Ben: Use case doc
... Guus review: should we mention RDFa?
... we modeled document after grddl

Guus: strictly speaking it is not a UC doc if you mention RDFa
... to avoid too much technology-driven document
... and overlap between primer and UC

Ben: OK

<mhausenblas> does this also effect the code snippets?

Ralph: it would be artificial

Guus: Parnas shows it was possible to rationalize after design

TimBL: it makes sense to explain the kind of things you want to do


Ralph: there are still things undecided
... eg. how much rdf/xml we want in rdf/html
... a use case can explain the boundaries we want to have

Guus: UC are useful for scoping
... explaining to the outside public our decision

Ben: there was already some consultation outside

TimBL: did you find case for publishing full RDF?
... or just a chunk?

Ben: there were cases (like bibtex) that caused us to rethink

TimBL: are there things we cannot do?

Ivan: problem of expressing lists/containers

<timbl> Tim: Is there a well-defined list about what can't be expressed?

Ivan: reification, but less important

<timbl> Tim: reification is negatively important IMHO

Ben: we have also datatypes
... action to take that list of exclusions of the wiki

<RalphS> ACTION: Ben start a list of RDF/XML features that are not supported by RDFa [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action01]

Mark: RDF community might want us to resolve their problems according to current best practices, e.g. for reification
... not clear which way we should go to the broader community
... I think Guus point on UC is relevant
... not having sample markup makes sense
... seems wrong that UC doc looks like primer

Ben: what is the WG opinion on removing rdfa code from doc?

WG: approves

<TomB> +1 on removing RDFa snippets from UC doc

Michael: link with microformats?

Guus: could be a good test

Ben: the requirements we have now are likely to go further than microformats

Ralph: it's ok to mention rdfa some times, to build a brand

Guus: removing the code is what is really needed

<mhausenblas> instead of removing RDFa code, why not ADD microformat code :)

Ivan: From the very start the goal was that rdf/xml could be fully embedded in html
... in some cases it proved to be too complicated
... like for reification
... UCs do not include reification now
... we could revisit that goal

Ben: we could annotate triples like provenance of license info

<Zakim> FabienG, you wanted to talk about GRDDL use cases doc

Fabien: there is no code is grddl use cases
... first part: problem we propose to address, no mention to grddl
... second part: how grddl could solve the problem

Alistair: RDFa primer is best doc to go for info

<Zakim> aliman, you wanted to say selling RDFa in use cases potentially wrong

<FabienG> GRDDL use case scenario doc for info : http://www.w3.org/2001/sw/grddl-wg/doc43/scenario-gallery.htm

Guus: I would prefer if RDFa design goals were not trying to address complete RDF
... it should be easy to understand, simple document
... if we spend one year on addressing everything we might propose something scary for the community

Mark: this was a flexible design goal, no criterion for success
... originally there were request like this document is about that with 80% certainty

<mhausenblas> +1 to Mark

Mark: there could be other ways that reification
... I argue against too much simplicity

Guus: there are conflicting requirements for any technology

<mhausenblas> we could also do a CFA ( as http://www.w3.org/2005/rules/wg/wiki/WD-DC)

<timbl> Mark seems to argue for information about triples, which to me suggests graph literals .. a way of putting a wrapper around some rdf/a.

Guus: we have to solve this req of simplicity while supporting all the use case
... there is no problem with having such conflicting req

<Zakim> MarkB_, you wanted to explain my outstanding action item on reification v. n-ary relationships.

<Zakim> timbl, you wanted to discuss collections

TimBL: drawing the line between what's in and out
... annotating triples is important
... RDF bag and sequences are difficult to work with
... some applications (creative commons) use it
... also valid for sequences

Ben: ol in html embodies collection, we try to see that

<timbl> ol a collection, ul a class maybe

<RalphS> PROPOSE: RDFa is not required to support every feature of RDF/XML

Ralph: we should apply the same process as for SKOS UCs yesterday
... we have to wait for a use case before deciding whether reification is in or not

<Zakim> RalphS, you wanted to propose a resolution re: RDF/XML completeness

<mhausenblas> http://www.w3.org/2005/rules/wg/wiki/UCR/Methodology

Michael: make sense to make a critical factor analysis on wiki?

Guus: it could be a good idea, it would take more time
... in Rule wg, use cases were much more difficult to analyse

<Zakim> kjetilk, you wanted to ask about v2

Kjetilk: we have good idea of what people want to do now
... and what people might want to do later
... could we introduce a version 1 and a version 2?

Ben: in theory, yes, in practice we would have to be really careful

<RalphS> PROPOSE: RDFa is not required to support every feature of RDF/XML

WG agrees

RESOLVED by consensus: RDFa is not required to support every feature of RDF/XML

use case 1: basic structured blogging

Ivan: for many people this might not be relevant: many bloggers do not use html

Ben: this use case means that people can write plug-ins to do that
... UC1 should be clearer about the tool support

Mark: there might be a use case with markup by hand

UC2: publishing an event

Ben: UC2: publishing an event
... could go to a specific tool (creative commons) to get a machine-readable chunk and copy-paste in html page
... UC1 could be for tool support for RDFa, UC2 more wizard-like

Ivan: one big feature of RDFa is mixing vocabularies
... if we look at this UC, microformats could do that
... event information should mix different vocabulary
... just some words to be added to the text

TimBL: put the RDF to include in the HTML in the example

Mark: other benefit is the use of existing taxonomies
... problem with microformat is that you have to reinvent taxonomies
... should we include two different use cases?

Ivan: additional benefit: author can add his own namespace

<RalphS> Ivan: mix events, bibtex, geolocation

UC3: content management metadata

Ben: use case 3: content management metadata
... various decisions about content
... the structured data may not be rendered

<mhausenblas> +1 to Guus suggestion

Ivan: this code seems to be xhtml2 perhaps not wisest thing to do
... also technical issue: if final design is only to add attributes or to change content model
... raise more problems if we want to combine with text

Ben: this is a fair comment, to take into account

use case 4: creative commons use case

Ben: self contained chunk added in html

<Zakim> Guus, you wanted to suggest that we explicitly discuss at the end of the document why MF are not sufficient for handling the use cases

Guus: section in the document where it is said that MF are not enough to solve the problems

Ralph: I think we do that by presenting use cases
... if MF meet challenges then MF are the solution

Mark: for the simple use case we could show that MF and RDFa can solve the problem

<mhausenblas> +1

Mark: and for complex ones, that only RDFa is OK

UC5: clipboard

Ben: comment by alistair this was not a distinct UC
... I think it is important, I can demo it

Guus: there is no problem with overlapping UCs

Alistair: copy-pasting in html has nothing with RDFA
... the point is that if I copy html with rdfa statement, I want them to be included when pasting

Ben: there should be a way to associate with a certain region of the interface some statements
... that should be copy-pasted and brought somewhere else

TimBL: the need is to copy-paste html with all the rdfa about this piece

<RalphS> TimBL: there's a sense of locality to the RDFa and HTML markup

Ben: should emphasize the need for localize relevant rdfa statements for copy-paste

<timbl> depends on the ability to localize the data to a part of the doc

UC6: semantic wiki

Ben: rdfa as input when editing wiki and having it in result

Ivan: is that really rdfa? there would be a different syntax

<FabienG> http://www.w3.org/TR/grddl-scenarios/#wiki_use_case

<Zakim> FabienG, you wanted to talk about GRDDL equivalent use case : http://www.w3.org/TR/grddl-scenarios/#wiki_use_case

Fabien: good idea to link with the GRDDL wiki UC
... lot of semantic wiki get rid of wikiML and just copy-paste
... wysiwyg interfaces are preferred

Michael: two issues

<mhausenblas> http://www.w3.org/2005/Incubator/mmsem/wiki/MichaelHausenblas

<FabienG> example of WYSIWYG interface using XHTML and RDFa for a wiki: http://argentera.inria.fr/wiki/data/Main/MainHome.jsp

Michael: requirement link to multimedia semantics WG
... using a wiki syntax related to rdfa

Ben: let's not focus on rdfa as input
... but you could paste rdfa

<Zakim> kjetilk, you wanted to ask about bbcode in foras

<mhausenblas> can you provide a pointer, please?

<FabienG> BBCode : http://en.wikipedia.org/wiki/BBCode

UC6: structured publishing by scientists

Ben: motivated by existing chemist blog
... UC is more advanced user agent, getting local RDF

Ivan: so emphasis is on adding some sexy UI to visualize the RDF info
... what I like is reference to other community
... you should put some more reference to science commons
... The difference here with MF is that vocabularies are huge

Mark: I agree with that
... perhaps the structured blogging UC should be different

<Zakim> RalphS, you wanted to note reference to browser enhancement

Ralph: this last UC mentions application-specific extensions

Guus: brainstorm with suggestions of applications

TimBL: 3 UCS
... 1: have RDF recording a collection of authors
... bibtex can be used as example, but point should be made that order should be kept
... 2: UC with unordered list: list of references for a WG
... owl:oneOf
... 3: UC: collect a foaf file manually done

<Zakim> FabienG, you wanted to list the GRDDL use cases in case one could be inspiring.

Fabien: 3 use cases from GRDDL


<RalphS> [[

<RalphS> # Use case #1 - Scheduling : Jane is trying to coordinate a meeting.

<RalphS> # Use case #2 - Health Care: Querying an XML-based clinical data using an standard ontology

<RalphS> # Use case #3 - Aggregating data: Stephan wants a synthetic review before buying a guitar.

<RalphS> # Use case #4 - Querying sites and digital libraries: DC4Plus Corp. wants to automate the publication of its electronic documents.

<RalphS> # Use case #5 - Wikis and e-learning: The Technical University of Marcilly decided to use wikis to foster knowledge exchanges between lecturers and students.

<RalphS> # Use case #6 - Voltaire wants to facilitate the extraction of transport semantics from an online form used to edit blog entries.

<RalphS> # Use case #7 - XML schema specifying a transformation: the OAI would like to be able to specify document licenses in their XML schema.

<RalphS> ]]

<RalphS> -- http://www.w3.org/TR/grddl-scenarios/

Fabien: relation between grddl and rdfa
... one use case is a counter-example
... case where it is explained that sometimes it can fail

<RalphS> Fabien: GRDDL UC editor's draft contains a new use case 8 counter-example

Jon: metadata registry which express the vocabularies
... we want to embed the RDF in HTML for rendering

Ben: could be interesting to have a SKOS-specific UC

Ralph: maybe an online dictionary can include some SKOS

<FabienG> Counter-example in GRDDL use cases current draft: http://www.w3.org/2001/sw/grddl-wg/doc43/scenario-gallery.htm#html_tidy_use_case

Guus: UCs having different vocabularies

<RalphS> Ralph: dictionary or our HTML wordnet files might include SKOS markup

Guus: food domain
... product catalog
... this is kind of UC which is not emphasized currently

Stephen: UC with retailers, venders, with multiple vocabulary. HTML view of last financial transactions in RDFa, interpretable by browsers
... trip organizer to help with decisions
... news stories, journals: grab all the key ideas about the stories you care about

Ben: comments on primer are editorial
... we can skip it

<FabienG> Counter-example in GRDDL use cases current draft: http://www.w3.org/2001/sw/grddl-wg/doc43/scenario-gallery.htm#html_tidy_use_case

GRDDL use case and RDFa use cases

Ben: GRDDL agent have an RDFa parser
... other option hGRDDL e.g.: transform microformat into RDFa
... this would preserve the locality in a new HTML doc

Ivan: third option use the GRDDL mechanism to extract RDFa

Alistair: GRDDL agent: does it have to parse it (RDFa parser) or does it use a GRDDL transform?

Guus: we must write down the relationship in the GRDDL doc and in the RDFa doc

<scribe> ACTION: Ben to write down the relation between GRDDL and RDFa [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action02]

Ben: RDFa would be a recommendation for an XHTML module

Fabien: one problem is that RDFa is presented as a new syntax for RDf whereas GRDDL is presented as a way to extract RDF/XML from other XML syntaxes
... a GRDDL transformation from RDFa to RDF/XML doesn't make a lot of sense to me if RDFa is adopted as an alternate RDF syntax
... an agent is either a GRDDL agent or an RDFa agent

Ben: this should be an GRDDL working group decision

<Zakim> timbl, you wanted to talk about the ladder of authority

Tim: explains ladder of authority
... we must decide if its part of HTML or if we use the GRDDL way

REC discussion

Guus: back to REC discussion

Ivan: if we produce an XHTML modul REC, we would need a new DTD
... module as a REC would not solve the validation problem
... the XHTML WG owns these DTDs

Ben: the validation would be separate

Tim: yes but other validators would complain

<Steven> anyone can create a driver

Ivan: the driver in the XHTML 1.1, a change has to be made and it is something this WG can't do

Steven: not such a big problem to make driver
... a document that wants to be validated has to reference the modified DTD

Ben: it would be good if we produce a validator as part of this WG output.

Mark: we are not modifying XHTML 1.1 and we can't.
... XHTML Modularization 1.1 is a different thing.

<Steven> XHTML1.1+RDFa

<timbl> unless we change the XHTML 1.1 DTD

<Steven> unless we create XHTML 1.2

Guus: issues wrt REC: resources for test cases, resources for team contact.
... two RECs may be too much work for this WG.
... do we have sufficient people to set up test suite?

<ivan> elias torres

Ben: Elias Torres from IBM would be of great help for test suite
... we have material for the tests we have to assemble them

Guus: set up a repository?

Ben: not too worry about that several people can help (Ben, Mark, Elias, etc.)

<RalphS> formal WG participants [Member-only link]

Guus: I am concerned about not having enough resources to make significant progress.
... is the schedule realistic.

Ben: agressive but we must do it if we want to have this done.

Ivan: what are the alternatives?

Guus: not going for Rec would be one alternative.
... then reconcider when we finish up

Ivan: resource shortage is the plague of the whole SW activity.
... Steven could help if he could spend some of his time on this issue.

Ralph: a lot of the work is editorial

Tim: since there is no more resources should we reconcider if we want to go with this?

Guus: we have to check internal depedencies, etc.

Tim: we must identify what a new WG resource would be doing precisely. What exactly should be done?

Ben: I prefer to take the risk to fail than to cancel it now.

Ivan: we have to have relatively stable publications on a regular basis for RDFa because there is a lot of controversy around it.

Ben: even if we don't reach a Rec we could stabilize a version as a Note.

Guus: that would be my proposal "go for REC track"

ACTION: Guus to flag the issue of RDFa REC track on the coordination group [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action03]

ACTION: Ben to get the docs in good shape for next week [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action04]

Tim: I am concerned about the fact that RDFa attributes semantics to an HTML doc.

<RalphS> I am on queue to respond to that concern

<MarkB_> I couldn't hear properly, but it sounded like there was a proposal for a discussion about whether an HTML document should 'flag up', whether it contains RDFa or not. I wrote a long email to the list about this, in response to Ivan, but no-one has commented on it. I would therefore appreciate it if no *final* decisions were taken on this issue at this meeting, since I won't be able to participate in the discussion.

Guus: break out sessions for this afternoon
... SKOS integration of issues and requirement list
... RDFa discussion on use cases, GRDDL relation, etc.

Principles for Managing Vocabularies


Tom: About Voc Management Note: ... we don't have an editor
...20050705 is old. Refer to the wiki version now.
Motivation is to describe what's involved in publishing an RDF voc
good to step back now and distinguish soft rec and hard rec
... and cite the cookbook
..principles of best practice may need to include very basic, pragmatic things such as "remember to pay your domain registration fees", etc.
List and describe basic advice, low-hanging fruit.
Today: brainstorm on the list of 5 points to see if it is a good starting point.

1. Name Terms using URI References

Tom: DanC said we should identify terms with URIs ; this is a hard REC for publishing an RDF Voc: http://lists.w3.org/Archives/Public/www-archive/2005Mar/0006.html

Ralph: Dan's mail was more a terminology point than an architecture point.

Tom: naming convention could be moved from the cookbook

Ralph: I wouldn't want to move it but there should be cross references.
... good place to mention the domain registration problem.

2. Provide readable documentation

Tom: the whole documentation question could be grouped in one point with pointers to the cookbook
...not giving too much details on what web pages one has to create to publish a voc

Ralph: considered best practice to have both human readable and machine readable doc
... we should show examples of what we think are best pratices.

Alistair: web page issue is "what a doc web page should look like" and show examples.

Tom: examples of different granularity in documenting

<TomB> Dublin Core documentation: http://dublincore.org/documents/dcmi-terms/ and http://dublincore.org/usage/terms/history/

Alan: a way to make this point would be to pose a little query pb and let people discover what can be done and what can't be.

3. Articulate Maintenance Policies

<RalphS> Ralph: seeAlso -> http://www.w3.org/1999/10/nsuri "URIs for W3C Namespaces" - W3C Namespace usage policies

Tom: there should be example of different types of voc and the maintenance policies that they have.

Ralph: section 3 in http://www.w3.org/1999/10/nsuri describes how namespace UI may change over time

Guus: what about the versioning?

Tom: we can show an example of URI used to identify snapshots of voc
... Dan Brickley is interested in using Web CVS to expose different versions of voc

Ralph: it would be good if we did propose ways to identify versions

Guus: I have pb to see difference between point 3 and point 4

4. Identify Versions

<JonP> Cookbook-related suggestion postponed from yesterday: http://www.w3.org/2006/07/SWD/wiki/BestPracticeRecipesIssues/ServingSnapshots

Tom: describing how versioning is done in large voc repositories and in SKOS is also relevant

Alan: hard part of versioning is to identify the policy people are using

Alan: term-level versioning vs. vocabulary-level versioning are choices people make
...would be a good thing to identify the possible policies
... ponters to implementation would also be interesting

<Elisa> Another source for metadata and examples regarding versioning policies is the BioPortal (from the National Center for Biomedical Ontology), at http://www.bioontology.org/ncbo/faces/index.xhtml

<Zakim> aliman, you wanted to ask about KWeb rdf versioning

Alistair: on KnowledgeWeb there are pointers to RDF versioning tools

Guus: that would be a possibility to have these people involved

<RalphS> Knowledge Web Network of Excellence project

<alanr> http://www.mail-archive.com/public-semweb-lifesci@w3.org/msg01624.html starts a thread about versioning.

ACTION: Guus to contact persons working on versioning in KnowledgeWeb [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action05]

<alanr> http://biopaxwiki.org/cgi-bin/moin.cgi/BioPAX_Level_2_feedback/BioPAX_namespace_issue

Tim: when you introduce a new namespace you can use OWL to publish the relationship between the old version and the new one.

<RalphS> yes, candidate Best Practice: use OWL [and some other vocabulary] to describe the relationship between any changes you make in your vocabulary to the previous version

Tim: I mean using sameAs, equivalent*, etc.

<alanr> http://ontology.buffalo.edu/bfo/Versioning.pdf

Tim: this is a real added value of RDF

<timbl> The TAG has been trying to deal with XML versioning and there is much less one can do in general.

Ralph: enumerating policies and examples of them would be a good added value already

<RalphS> Jon's proposed COOKBOOK-I3.1 issue

Jon: taking a very concrete example is teaching people how to cook the cake and not how to read the cookbook

<RalphS> Jon: I believe this versioning discussion subsumes COOKBOOK-I3.1

Guus: I'd like to see a practical of example e.g. in the medical domain

5. Publish a Formal Schema

Tom: give good practice of how voc are being declared would be enough without going into too much details

<kjetilk> http://my.opera.com/community/xmlns/2006/gallery.rdf

<alanr> versioning perhaps http://www3.lehigh.edu/images/userImages/jgs2/Page_3813/LU-CSE-06-026.pdf

Kjetilk: don't we need a voc to describe our mainteance policy?

<RalphS> we should say that it is best practice to publish an RDF/OWL document at the namespace URI. This may be obvious to us but evidently it's not obvious to everyone. Point back to Recipes document for "... and here's how"

Guus: it is a good idea in principle. But we are not doing new work here we just identify existing practices.

Alistair: what are the plans for the semantic interop note.

Guus: subject for last session.
... looking for an editor: Elisa ? Kjetik ? etc. ?

<TomB> I can contribute descriptions of how things are done with Dublin Core.

Elisa: Daniel could also contribute with examples

ACTION: Elisa to give first overview of what the status of the doc is and add comments and coordinate work on doc [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action06]

Breakout on SKOS

guus: revisit all issues we discussed, identify candidate requirements, bring us to position of having first complete list of requirements, useful?
...finish quickly, can look at remaining use cases

Antoine: jon updated requirements list yesterd, based on that I created a bullet list of the issues

<Antoine> http://www.w3.org/2006/07/SWD/wiki/SkosIssuesSandbox

<aliman> candidate requirements sandbox

<aliman> antoine: wanted to collect stuff from yesterday, from old issues etc. perhaps more issues than what we need

guus: propose to split requirements into candidate and accepted
... have the notion of relations between string values
... in requirements, have notion of acronym, representation of realtionships between labels associated with concepts

Annotations on lexical labels

Guus:... making statements about lexical labels? alan, boil down to ability to represent statements about lexical labels?

alan: yes. and the choice between boosting labels to individuals and using Alistair's pattern (n-ary relations)
... real example from obi, this term is used by community x, this term was proposed by x, needs to be reviewd, was reviewd on x, this term was in use 200-500 b.c.

guus: statements that relate a lexical label to a resource, and to various data values, typed data values like timestamps.

alan: yes, the resource might be the container, depends on representation choice. other way is if lexical item is an individual, properties hang off it.

guus: one way to handle this is to reformulate requirement 3, or add new requirement. currently req 3 is acronym example. boils down to same thing, make statements about things ...

jon: talking about metadata?

guus: from representation perspective same problem, from use perspective it is different.

alan: synonymy is relationship between terms, like acronym example. could be considered same issue.

guus: prefer to have separate requirements for now. name?

alan: annotations on lexical items, how to represent?

guus: requirement should be, the ability to represent annotations on lexical items.

antoine: want to control the actions, e.g. for this issue alan has two actions, one is to write down the general documentation requirements and how to represent in SKOS, then another item to write up preferred label modelling issue

aliman: what is preferred label modelling issue?

antoine: for me it was this issue of lexical values and annotations

aliman: let's get rid of "prefLabel", misleading

antoine: i'll change


guus: next issue ... MappingToCombination

antoine: I added this one, but no reference in the minutes for action

guus: conjuecture, have separate req on compositionality, req 8th in list reflects the issue
... issues are things where we have to propose a resolution on how to this, lead to test cases, if you can't find any req to which an issue refers there is something wrong. based on issues, are there any missing requirements?

guus: specialisation of relationships, we have a req for this ... local specialization of SKOS vocabulary - so this is covered.
... relationships between labels, we have this one covered by the req mentioned before (number R3)
... next set of issues more tricky, because there should be use cases if we admit as requirement
... (now looking at issue SKOS-I-Rules)
... need to think of motivating use case where need rule

antoine: manuscripts use cases or any use case where propagate indexing up hierarchy levels

aliman: SWED use case uses this rule

alan: can do in OWL 1.1 role inclusion

aliman: we're not waiting for owl 1.1

alan: yes but good to be aware

guus: same thing as checking consistency?

aliman: no, more about inferring new information

guus: SWRL document, first example has a rule like, relationships between artists and styles, can derive the relationship.

<alanr> http://www.cs.man.ac.uk/~horrocks/DAML/Rules/

antoine: rule [from SWRL] more complicated than indexing example but similar

guus: have anumber of use cases, what is formulation of requirement?
... not talking about part of skos specification. what could be part of skos specification is bt and nt are inverse of each other, rt is symmetric, bt nt transitive?
... originally sean mentioned this issue because ruiles in skos
... doesn't currently give rise to requirement, might if we resolve this.
... 2.1.2. SKOS-I-ConceptSchemesContainment . ..
... the semantics of containment withing particular vocabulary./ontology is not clear.

steve: is this also strictly contains/is contained by, or is alan calculus, all different modalities of containment, connection, proximity, applies for temporality and spatiality ...

alan: technical issue, can hook up a concept to a concept scheme via a property, but can't do the same for a triple

antoine: motivation is that a given concept e.g. france, might be narrower different things in different concept schemes

steve: problem is e.g. with gazetteers, often strict containment e.g. DC area overlaps with other areas
... so next step beyond strict containment that talks about whether things are next to each other.

antoine: more about reification, statements made in the context of specific concept scheme

aliman: concrete use case?

alan: case came up yesterday, equivalence from one point of view

guus: hesitant on this issue, goes beyond level of RDF OWL, e.g. look at RDF/OWL ontologies, containment is implied by containment in files, i.e. informally, but if RDF OWL didn't give any semantics to that, why do it in SKOS?

alan: because if needed in this domain then yes.
... reify, don't use RDF reification, promote relationships to individuals, can describe any properties of the relationship
... e.g. look at a mapping [draws on white board]

how do point a statement to the containing scheme

scribe: something close to that RDF reification

guus: suggest we post a candidate req, have an explicit representation of the containment of concepts or relations
... any element of a concept scheme (could include concepts, relationships) ... rdf/owl soolution is implicit
... have to make this explicit

alan: requirement would be, relationships need to be explicitliy associated with scheme, also concepts

guus: also specify for concepts, for both there is a requirement
... good to know if it's part of original vocab, or if someone added it

alan: understanding was, asymettric equivalence, can't do without ???

guus: almost all tools have way to ask [SPAARQL] can ask the database question and the logical question, which is two different things
... e.g. can ask direct subclass of, does it eist, or has it been inferred?

alan: direct vs. indirect different from told vs. inverred

antoine: really close i think to asserted vs. inferred

guus: could be that this is solved at query level and not at representation level

alan: may have inferred intervening class, therefore problems are not quite same

antoine: at dutch library which has broader links which are redundant,
... e.g. asserted closure

steve: get from forward chaining ...

guus: candidate requirement for the moment, can always disregard

the ability to explicitly represent the containment of any individual which is an instance of a SKOS class (e.g. skos:Concept) or statement that uses SKOS property as predicate (e.g. skos:broader) within a concept scheme

the ability to explicitly represent the containment of any individual which is an instance of a SKOS class (e.g. skos:Concept) or statement that uses SKOS property as predicate (e.g. skos:broader) within a concept scheme

guus: understand by now then happy
... issues from previous SKOS issues list ... collections-5 ... fix expression of disjointness between concepts and collections
... doesn't generate representationrequirement
... do we have policy on using SKOS namespace for something not in SKOS ... issue in OWL, parsers should flag but otherwise continue as normally, triples using bad URIs get no semantics

<TomB> skos:ConceptScheme as a set of concepts, or specifically skos:Concepts?

guus: allow people to use extensions that later on

tomb: also ability to extend SKOS, e.g. other types of concept
... other class in other namespaces then are we covered?

aliman: always subclass skos:Concept

tomb: what can a concept scheme contain? limited by class skos:Concept? should that be stated somewhere?

guus: raise general issue on how to represent SKOS semantics, not at all trivial, good feeling of what semantics should be, but how to represent is another thing

<scribe> ACTION: alistair to raise a new issue about USE X + Y and USE X OR Y [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action07]

guus: metaphor aliman just gave between descriptor and non-descriptor is excellent to help explain what these things mean, really from ... just want to have same term on a card.

alan: explain what indexing meant, explain descriptors and non-descriptors
... guus showed in demo groupings of terms that were not terms
... in final session talk about scheduling

Discussion and wrap-up

The RDFa breakout session

The wiki is uptodate but is being transferred to tracker

ACTION: Ben to update issues list with the @CLASS overload problem [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action08]

did not get to the GRDDL issues

did not discuss planning

Guus: we should decide REC or NOTE in april
... before the summer, we should have last call in the case of REC
... we could ask for CR by october
... that would get us to REC by the end of the charter
...there should be one WD before the last call

benadida: we should have the WD just before the REC decision

Guus: we need to pay close attention to the outside world

RalphS: we don't know the status of a XHTML 2.0 WG

ACTION: Ben to update RDFa schedule on wiki to aim for last call on June 1 [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action09]

The SKOS breakout session

aliman: we went through a sandbox list, also included some new requirements: SKOS Requirements List Sandbox

aliman: we had a long and philosophical discussion of the wording of point 22

Guus: what would be a reasonable schedule for coming up with a first WD?
... it doesn't need to be complete
... just useful for review
... by march would be realistic

Antoine: yes, it sounds doable

aliman: we just have a primer and a formal spec and a reference overview document

aliman: the problem with the guide is that it does two things, like give an introduction to SKOS as well as defining some of the semantics

<alanr> http://www.zuser.org/z/

Alistair: I used Z for the formal specification language for my thesis

ivan: I sweated a lot over Z

aliman: we have a few high-profile users of thesauri that are involved

ACTION: aliman to update the schedule for SKOS documents [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action10]

Summary of Action Items

[NEW] ACTION: aliman to update the schedule for SKOS documents [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action10]
[NEW] ACTION: alistair to raise a new issue about USE X + Y and USE X OR Y [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action07]
[NEW] ACTION: Ben start a list of RDF/XML features that are not supported by RDFa [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action01]
[NEW] ACTION: Ben to get the docs in good shape for next week [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action04]
[NEW] ACTION: Ben to update issues list with the @CLASS overload problem [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action08]
[NEW] ACTION: Ben to update RDFa schedule on wiki to aim for last call on June 1 [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action09]
[NEW] ACTION: Ben to write down the relation between GRDDL and RDFa [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action02]
[NEW] ACTION: Elisa to give first overview of what the status of the doc is and add comments and coordinate work on doc [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action06]
[NEW] ACTION: Guss to flag the issue of RDFa REC track on the coordination group [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action03]
[NEW] ACTION: Guus to contact persons working on versioning in KnowledgeWeb [recorded in http://www.w3.org/2007/01/23-swd-minutes.html#action05]
[End of minutes]