Teleconference.2008.03.05/Minutes

(Scribe changed to Vipul Kashyap)

Alan Ruttenberg: Rearrange agenda to get the fragments discussion earlier

Achille Fokoue: Start the discussion on fragments earlier

PROPOSED: approve previous minutes

RESOLVED: approved minutes http://www.w3.org/2007/OWL/wiki/Teleconference.2008.02.27/Minutes

Alan Ruttenberg: Issue 95 discussion to resolve... Was the action item completed?

Alan Ruttenberg: How to resolve Issue 3, Tabled for now

Alan Ruttenberg: seek feedback on Action IItems

Alan Ruttenberg: Close pending action items

Alan Ruttenberg: Status of Action 86?

Alan Ruttenberg: Jeremy to respond?

Alan Ruttenberg: Issue 97, Adding GRDDL to the OWL XML syntax.

Alan Ruttenberg: Transform OWL-XML into OWL-RDF

Alan Ruttenberg: Do we accept this issue?

Bijan Parsia: OK with accepting the issue

Ivan Herman: Each XML dialect has a namespace

Ivan Herman: Describes how GRDDL transformations are identified and invoked using the GRDDL standard

Alan Ruttenberg: Schedule a discussion on the issue

Alan Ruttenberg: Issue 96 on next week's agenda

Alan Ruttenberg: Issue 95 to remove datatype restrictions to say things like complementOf?

Alan Ruttenberg: remove nested restrictions on datatype range

Alan Ruttenberg: Resolve this issue as is and add a new issue to name and further restrict datatypes

Alan Ruttenberg: Start with Fragments Discussion

Alan Ruttenberg: Pragmatics of organizing work going forward

Alan Ruttenberg: 3 fragments

Alan Ruttenberg: Overlap between OWL Lite and OWL Prime

Alan Ruttenberg: Overlap between DL Lite and OWL Prime

Alan Ruttenberg: Task Force devoted to each Fragment

Peter Patel-Schneider: How much work needs to be done, given the work done by Bors and Bernardo?

James Hendler: Agree with Peter... Work of this document or have another task force?

James Hendler: Far enough ahead, so probably put it in as a document of the workgroup?

Alan Ruttenberg: Discuss document on the mailing list and work through issues that come up on the mailing list

Sandro Hawke: Is this a Rec track document?

Alan Ruttenberg: Replace the document on the wiki?

Ian Horrocks: Replace the existing fragments document with the new one on the premise rec track?

Ian Horrocks: What do you do with the existing one?

Jeremy Carroll: In favor rec tracking this document

Ivan Herman: Favor to rec track

Ivan Herman: Do not have to do this decision at the moment

Ivan Herman: Turning the submission 1 into a note might make sense, but do not need to decide on this right now

Bijan Parsia: Agree with Ivan, Nice Template to describe fragments

Bijan Parsia: Making this docuiment nice to read would be good

James Hendler: Current document should be somewhere accessible...

Alan Ruttenberg: Leave the document "tractable fragments" remain where it is and change the draft pointer to the new location

James Hendler: Need ability to look at both of them carefully

Ivan Herman: OWL 1.1 input submission links to Fragments document. 2 links away from wiki page... Don't need to do anything

ACTION: Alan to change sidebar Fragments link to new page

James Hendler: Entry recognized W3C format

James Hendler: built rules for the fragment in that

James Hendler: It is an executable format as engines can execute N3 descriptions

Bijan Parsia: Separate what format vs how to coordinate/reference other W3C work such as RIF

James Hendler: N3 representation of the rules is useful and also gives mapping to RDF

Sandro Hawke: Wait for the RIF last call

James Hendler: General discussion - What flavors of presentation of syntaxes etc are we going to use?

James Hendler: If we use different syntaxes - all syntaxes should be used in other documents

James Hendler: ... use of standard rule formats

Sandro Hawke: N3 Rules do not have stable documentation

James Hendler: some of it may be in the N3 document

Alan Ruttenberg: Separate issues related to syntax versus putting the rules on wiki

ACTION: Hendler to put n3 version of rules on wiki with pointer to documentation. All to review and discuss via email

Alan Ruttenberg: Publication schedule: 3 documents - 3rd document is fragments document

Michael Schneider: For what applications is the XML doc needed?

Michael Smith: Implementation experience with XML format within DIG client/server architecture for passing explanation in the context of an NCI project

Bijan Parsia: Matthew used it in MyGRid experiments

James Hendler: Useful, but confusion on status of various documents

James Hendler: If published as a draft , can decide whether to stop it as a note or go to a rec?

James Hendler: XML exchange syntax not to be published as rec

Ivan Herman: What are the arguments for the above and what has changed since then?

Scribe not: Lines -1, -2 by Ivan and not Jim

Jeremy, Peter: If we got to rec, we would have to look for implementations

Alan Ruttenberg: Things in XML you cannot do in RDF/XML - undue burden on RDF/XML clients

Alan Ruttenberg: Validate: GRDDL transformation can be written to correct the syntax errors

Boris Motik: DIG simple protocol for DL reasoners, committed to use XML syntax

Boris Motik: DIG is using XML syntax for updates ot KB, DIG is axiom based

Bijan Parsia: People in XML groups do not like RDF/XML

Bijan Parsia: We are thought as XML hostile

Bijan Parsia: Have a transformation to RDF (not document)

Bijan Parsia: Opens OWL to people who might be turned off by RDF/XML syntax

Peter Patel-Schneider: Task for XML to RDF is impossible

Peter Patel-Schneider: cannot transform everything

Ian Horrocks: GRDDL transformation to check if we can go from XML to RDF.

Ian Horrocks: Can map from structural syntax to RDF

Alan Ruttenberg: sandro, please chair the rest of the session

Jeremy Carroll: Agree with Peter

Alan Ruttenberg: If I get OWL-XML which my OWL tools can grok, then I will have to retool. XML document should remain in track with them

Alan Ruttenberg: GRDDL transformation to keep in track and promote interoperability

Ivan Herman: understand the argument, Many things? N3, RDF/XML, OWL/XML, Functional synta

Zhe Wu: Conditions of some triples not convertable into RDF

Peter Patel-Schneider: WG has list of issues related to transformations into RDF/XML

Alan Ruttenberg: Predicates => Qnames => Start with digit character.... URL...

Alan Ruttenberg: Minor restrictions and do not get in the way

(Scribe changed to Rinke Hoekstra)

Jeremy Carroll: In RDF core, we considered the predicate problem Zhe was alluding to. It's an RDF problem, not an OWL problem...

Jeremy Carroll: we decided not to fix it

Jeremy Carroll: I understand alan, but am against it. But that wasn't what RDF core decided

Alan Ruttenberg: we have to live with the fact that RDF/XML might not be able to live with OWL/XML syntax

Alan Ruttenberg: don't want anything in the OWL/XML syntax that would not be compatible with RDF/XML

Bijan Parsia: is it bad to ask for new features?

Bijan Parsia: it is perfectly possible to push out some RDF that cannot be serialised into rdf/xml, but is valid RDF (turtle)

Bijan Parsia: it wouldn't be so difficult to provide a schema that would be able to check for compatibility with RDF/XML

Bijan Parsia: people should be allowed to write these things, and people already do this

Sandro Hawke: chairs will have to think about this next week

Issues discussions

Alan Ruttenberg: We might consider Issue 69 as a 'larger' issue, and spend some more time on it.

Alan Ruttenberg: if we have more time

Issue 91

Alan Ruttenberg: ontology properties, Issue 91

Alan Ruttenberg: boris added the ontology properties to the docs... but ontology properties have Ontologies as their range

Alan Ruttenberg: and domain

Alan Ruttenberg: that's where we are right now

Jeremy Carroll: I am still suffering HP overload. Domain and range can be specified in OWL full, but don't know how this works out in the FS

Jeremy Carroll: it's a finite list, so we could add them to the mapping rules (if necessary)

Jeremy Carroll: if someone could take this up, I'd be happy (busy until easter)

Michael Schneider: I assume here that the ontology properties will be in the RDF syntax, but in OWL DL they will be interpreted as annotations

Michael Schneider: but if they are in the RDF syntax they have automatically a domain and range in OWL Full

Michael Schneider: I wouldn't change anything in comparison to OWL-1.0-Full, where ontology properties have domain and range owl:Ontology

Peter Patel-Schneider: it's not really the case that they have a domain and range in OWL DL

Peter Patel-Schneider: (1.0) it was not something you could talk about in OWL DL

Peter Patel-Schneider: in a certain sense, historically this discussion is incorrect... don't think that anything extra needs to be done on these properties that is any different from annotation properties

Alan Ruttenberg: there have been various discussions on versioning issues. It would be desirable to have versions have more actual impact than be rather advisory

Alan Ruttenberg: have a proposal that I will bring up when we have our versioning discussion.... for the moment what I'm hearing is ok, but would like to at least discuss whether they could in some sense 'signal' something when something is 'bad'

Jeremy Carroll: answer to pfps, if I recall correctly, at least syntactically the range and domain have to be there, could be an issue with punning

Alan Ruttenberg: propose to leave it as is for OWL DL except to the extent that we have to say something about how it relates to Full.... for full we could followup on Michaels' issues

Alan Ruttenberg: suggest to close the issue with the note that the 'issue' (or similar) will re-arise in the future

Peter Patel-Schneider: to respond to jim in advance, the issue

Peter Patel-Schneider: under consideration says the spec lacks ontology properties. It doesn't anymore... that is different from the domain and range of said ontology properties

James Hendler: I disagree with you peter, but we're up to close to a 100 issues, just closing this to open up another one doesn't seem to help

James Hendler: we can put the issue off for a long time

Alan Ruttenberg: don't see any harm in doing as peter suggest, with the blabla to at least send an email about why and how

Alan Ruttenberg: strong objections?

Alan Ruttenberg: no reason not to keep it open, just send an email so that we know where we are

ACTION: Alan to summarize current state of issue 91

Issue 16

Alan Ruttenberg: Issue 16

Alan Ruttenberg: entity annotations status

Alan Ruttenberg: raised by jeremy

Alan Ruttenberg: defer?

Alan Ruttenberg: ok, let's defer Issue 16

Issue 90

Alan Ruttenberg: Issue 90, spec lacks deprecated marker

Peter Patel-Schneider: no change, specs have not changed...

Peter Patel-Schneider: proposed myself to deprecate deprecation

Alan Ruttenberg: heard some objections related to backwards compatibility

Peter Patel-Schneider: deprecation in the sense 'simply not say anything about it'

Alan Ruttenberg: wouldn't that lead to syntax errors?

Alan Ruttenberg: deprecation would mean 'we keep it here as a courtesy, but expect it to be gone in owl 2.0'

James Hendler: I sort of like to see a package on the whole issue of 'non-semantic' elements properties classes

James Hendler: have a confusion on the status of things we don't comment on

James Hendler: do these still apply or not?

Alan Ruttenberg: meta comment... the authoritative wording on any of this are the specifications. To the extent that you are concerned about this... read the specs and bring up the issues one at a time

Alan Ruttenberg: stop reading the emails (to this purpose)... only look at the spec

Bijan Parsia: I think I support the 'deprecation' in the sense that we say that they are deprecated (i.e. they were not sufficiently defined for the purpose of implementation in applications, we leave them in the notes and will probably come up with something that superseeds them)

James Hendler: object to any such resolution, see them as valuable in many applications.

James Hendler: leave them as is

Bijan Parsia: if I want to do deprecation, I cannot resort to the specs, because they don't specify it properly. We won't achieve compatibility across applications

James Hendler: there's a big difference between saying what this does, and specifying what it does. For humans2humans it works

Bijan Parsia: I'm not talking about formalization, only specification. would you expect it to affect the class tree?

James Hendler: no, don't expect it to, but in a future version

Alan Ruttenberg: you can simply read it as owlClass... the person reading it may take note that the class may be removed

Alan Ruttenberg: what would the harm be in leaving them in there, and provide better documentation

Peter Patel-Schneider: I would not see a problem if such declarations did not carry semantic weight in RDF. Unfortunately they do, and therefore I strongly believe that they need to specced much better than wishy washy human consumption only stuff

Peter Patel-Schneider: If one augmented the owl spec with that, then everything becomes hunky dory

Michael Schneider: they don't have specific semantics in OWL full apart from being classes

Alan Ruttenberg: might be a reasonable and not incompatible repair to add in the owl full semantics that owlDeprecatedClass sameAs Class

James Hendler: have a separate fix... would want them to be some kind of annotation

Alan Ruttenberg: could you work with michael on the owl full semantics to see how this would work out

Alan Ruttenberg: and see how this works out wrt peter's proposal

Alan Ruttenberg: meeting can be considered closed

ACTION: Hendler to work with Michael to clarify semantics of deprecatedclass so that peter becomes happy