SWD WG

 

 

<RalphS> Previous: 2006-10-17 http://www.w3.org/2006/10/17-swd-minutes.html

<TomB> Agenda for today: http://lists.w3.org/Archives/Public/public-swd-wg/2006Oct/0047.html

Admin

TomB: accept minutes previous meeting

scribe: minutes accepted

TomB: action on scribe conventions, next week

SKOS requirements

Sean: description of Cohse project
... using owl ontologies to find documents associated to concepts
... simple NLP to find expressions matching terms of vocabularies
... providing links between docs and concepts
... using ontology structure (hierarchy) to help navigation
... we have realized that using OWL ontologies for navigation navigation is not optimal
... thesaurus structure (broader/narrower) would be better

-> Sean's use cases

Sean: so replacing skos-like vocabularies, wrapping them up with simple services
... like synonyms, relationship retrieval would be useful.
... This use case fits pattern A: use vocabularary for retrieval, but also query the voc itself
... not necessarily a full RDF repository
... using a SKOS-like scheme on top would be relevant

Bernard: I am confused by the word "index", rather use "annotation"

RalphS: agree with Bernard re: discomfort with "index" - not sure "annotation" solves the problem... BT/NT jumps out as big benefit - easy to explain to people.

Bernard: Are there other examples of uses of relations, than broader/narrower?

Sean: related term could be useful

Bernard: less-precise BT/NT much better fit to the relationships we want to use for navigation than RDFS/OWL subClass

Sean: they are useful to access thesaurus

TomB: discussion should be postponed until we are more numerous

Sean: Question: is my document appropriate for a use case description?

TomB: yes

<TomB> Antoine: Qst about use cases.

<TomB> ...Can we provide abstract use case?

<TomB> ...Should the data referred to in the use case be public?

<TomB> RalphS: We should consider real-example use cases where data is not necessarily available. Like the level of detail of Sean's document.

<TomB> Bernard: SUN Microsystems ontology is not available - one of the use cases is Unified Product Taxonomy. Others are public-domain.

<TomB> RalphS: could you describe that in general terms?

<TomB> Bernard: When project started, there was a description - details are not public.

<TomB> RalphS: Level of detail we need to drive design of SKOS - we can do that without having...

Daniel : some stuff mentions in the use case can be published.

... it is actually important that some data is published!

<RalphS> Daniel's use cases

Daniel: terminologies were developped that could be used in computer services
... huge effort of knowledge creation in these cases
... goal is too create a large library of termonologies
... accessible for integration in various applications

Sean: one selling point of skos: semantic relationships to tie together concepts in a loose way
... is it relevant for the biology cases?

Daniel: tension btw terminologies (loose) and ontologies (rigorous, formal) - there is no easy answer re: best solution - if SKOS can support both communities that would be ideal.

Daniel: tradeoff between precision and recall - SKOS not appealing if doesn't accommodate need of some for crisp, formal semantics

Daniel: BT/NT very high-level - need more specific types of relationships
... more useful for reasoning services

Daniel: in anatomic models there are loosely defined links
... important communites are looking forward to adopt standards like dublin core and skos
... communities for functional genomics - consortium of researchers - basic biological research data - high-level ontology to subsume experimental investigation.

Ralph: Daniel, the first two cases would be interested in more precise semantics. Are there specific applications?
... what would motivate the need for more precise semantics?

Daniel: some communities are using classification to recognize inconsistencies
... they want to create more 'intelligent' services
... e.g. reasoning application which uses FMA

Ralph: what is the level of eagerness of these communities to follow swd schedule?

... what is our time window of opportunity?

Daniel: they want standards
... we're doing their work so timing is great

TomB: acknowledge Sean and Daniel

SeanB: Is there a gray area between scruffy and neat? some migration path
... from application neeeding simple knowledge
... to application needing rich formal semantics

Sean: Even uses of subsomption can help

Ralph: it might be interesting to extend on Daniel's examples refering to the SKOS draft
... use cases should point to things that work well
... and to things that need more attention

<scribe> ACTION: Daniel to link his use case to SKOS draft [recorded in http://www.w3.org/2006/10/24-swd-minutes.html#action01]

RDF in XHTML

Ben: everyone interesting in RDFs is invited to participate

<RalphS> meeting record: 2006-10-23 RDF-in-XHTML TF telecon

Ben: goal is to make RDF work as XHTML1.1 module
... 2 documents: RDFa syntax (more formal) and RDFa primer

<RalphS> editor's draft RDFa syntax document

<RalphS> editor's draft RDFa primer

... create chunk of html self-contained wrt metadata
... next steps: reification + containers
... target is 4/6 weeks for next working drafts

Ben:Many people would not want to get involved in huge level of technical detail

... but continuing copying important documents is useful
... to get feedback

RalphS: additional information that would be useful to have with SKOS use cases and schedule expectations for UC&R

Ben: about the documents issued by last TF
... they should get feedback from the WG

RalphS: an HTML WG shares these work items
... the charter for that group is still underway
... there could be some re-scheduling to match

RalphS:could some of Daniel's use cases both apply to SKOS and RDFa?
...Involving HTML documents with embedded metadata?

Daniel: could be. No immediate example now

Recipes for Publishing RDF Vocabularies

-> Postponed

<RalphS> Recipes Working Draft - comments from May 2006 [TomB 2006=10-24]