W3C

Semantic Annotations for WSDL WG teleconference

18 Apr 2006

Agenda

See also: IRC log

Attendees

Present
RA, CB, JF, MK, JK, EP, BNS, AS, John_Miller
Regrets
LH, HL, JBD
Chair
JacekK
Scribe
Rama Akkiraju

Contents


Roll call, agenda review

<JacekK> scribe: Rama Akkiraju

<JacekK> scribenick: rakkiraj

Jacek: John Miller from UGA may be joining us as an invited expert
... Invited experts role only for 6 months
... will think about John being an editor (for which he volunteers)

Approval of minutes

<JacekK> http://www.w3.org/2002/ws/sawsdl/minutes/20060411

Jacek: Last week's minutes: Approved

RESOLUTION: minutes approved

Action Items

Jacek: One action item from last week. Branhmananda has cvs account. The doc. hasn't yet started

<caribou> ACTION: BNS, JacekK and staff contacts to start the Examples document [ONGOING]

Jacek: can start the doc as soon as we see some examples on the public/private forum

First F2F

Jacek: Face-2-Face: DERI Galway will host 1st f2f.
... 13th-14th of June.

Mattias: Can't make tuesday and wed
... No, sorry, the dates are ok.

Jacek: Meeting logistics page will be setup ASAP.
... Date and place for first f2f is fixed as DERI galway.

RESOLUTION: first face-to-face in Galway, Ireland, June 13 and 14

Document status

Jacek: two issues are submitted. Will talk about it. If you have any comments submit them to the public mailing list.
... Title to be changed to 'Semantic Annotations for WSDL'

How about WSDL 1.1 Vs WSDL 2.0. We should combine it.

Joel: The current spec has a section on WSDL 1.1.

Jacek: We can keep it that way until it grows or if issues arise

Eric: supports Jacek's suggestion

RESOLUTION: Title changed to 'Semantic Annotations for WSDL'

rakkiraju: can post some examples soon after checking confidentiality further

New issues

Jacek: Moving onto issues

<JacekK> http://www.w3.org/2002/ws/sawsdl/issues/#open

Jacek: type of modelReference attribute

Joel: intention was URI, inconsistencies are errors

<ericP> +1 for URI from the RDF geeks

rakkiraj: A list of URIs would be good to have

eric: Would like to see usecases and scenarios

Jacek: May be we want to think about it a bit more

Joel: Multiple URIs seems good. But we need to discuss about how tools deal with these multiple URIs. Want to avoid complex syntax.
... How do we know what kind of a model it is?
... We need to understand what is required by tools to understand the type of the model

Jacek: Why not queue name? - Amit's question.
... Semantic languages all use URI's to point to things. Qnames are typed.

Amit: What we are adding to is WSDL. Why not keep it consistent with WSDL and think about QName?

Eric: Identify the semantics of agien object: Give a URI.
... semantics have diff. representations in diff. languages.
... IRIs allows us infinite extensibility with no pain.
... Extensibility gets confusing with QName

Amit: Looking for simple things from a developer's point of view.

Eric: Have to look at a usecase.
... What makes it easy to consume?

Amit: User will be driven by the ontology driven discovery process. Ex: Match a PIP in Rosettanet

Jacek: OWL, WSMO uses URIS
... UML uses strings. Does any of these models uses QNames?
... If none of them use QNamees, then we should just go with URIs.

Eric: Browsers use URIs too

Amit: can you always generate URI from a QName?
... If you do have QNames could you get URIs from it?
... Can they coexist?

Jacek: There was a proposal to transform QNames to URIs
... No proposal known to transform URIs to QNames

Amit: So, then if we adopt URIs we can't support QNames

Jacek: Languages that use QNames do also define URIs for identifying components.

Amit: Other side: Want to coexist with WSDL-RDF mapping.
... So there is an argument in favor of URI.

Jacek: There is preference for URIs.

Eric: Evaluate when it is an issue.
... When the person consuming the WSDL wants to do something with the QName.

Jacek: QName is more readable

Amit: What about a way to support both

Eric: if you hae semantics for interpreting QNames in RDF XML you concatenate the ns. You end up with something readable but still uses URIS

Jacek: Is there a mechanism to support both?
... We can't have both in unencoded form.
... Every QName is a URI where the prefix is interpreted as a URI scheme.

Eric: Syntactically differentiate them if you support both
... if you offer app level semantics to QName smeantics, you end with data types with QName but data model interpretation is that it is a URI

Jacek: Amit means something different. Clarification.
... Would like someone to come up wth a proposal. A scenario of why we need QNames

John: We had both URIs and QNames.
... in previous examples.

Jacek: Parser won't have problems parsing QNames ... but you loose the prefix-to-ns mapping.

Amit: Have a post on the mailing list and come back to it next week.

Eric: Convenience: Syntax or readability?

Amit: Mostly readability

Jacek: when using base and relative URIs, they are short and very readable

<scribe> ACTION: John Miller (responsible: Amit) to start off the discussion about issue #1 by looking at the implementation and seeing if base URIs and relative URIs also help the way QNames do.

Jacek: Closes Issue 1 for this week
... Issue #2: Embedded semantic descriptions
... There have been comments on SAWSDL/WSDL-S supporting embedded semantic descriptions.
... Either in a separate container or even in-place.

<ericP> +1

rakkiraj: worth discussing

John: It might make sense to have tags. Things could be useful.

Caribou: some people in the IG talked about that.
... So we should ask them to contribute the usecases

<scribe> ACTION: Carine to ask the SWS-IG about use cases for embedding semantic annotations

Jacek: It is worth discussing. Not much time left to discuss this.

Eric: Can show pointer to embedded descriptions

<ericP> http://www.w3.org/2005/11/SPDL/#amazonSearch

Jacek: Adjourn now. Think about these things and discuss them on mailing lists
... External annotations without modifying the WSDL is out of scope.

Amit: Can we pursue that subsequently.

Jacek: Changing of WSDL is a valid concern. But this is out of scope for this. Raise on the public forum.

Amit: may be useful for complex mappings and functions

<JacekK> meeting adjourned

Summary of Action Items

[NEW] ACTION: Carine to ask the SWS-IG about use cases for embedding semantic annotations
[NEW] ACTION: John Miller (responsible: Amit) to start off the discussion about issue #1 by looking at the implementation and seeing if base URIs and relative URIs also help the way QNames do.
 
[PENDING] ACTION: BNS, JacekK and staff contacts to start the Examples document