Teleconference.2008.02.13/Minutes

From OWL
Revision as of 18:09, 27 February 2008 by Sandro Hawke (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

These minutes have been approved by the Working Group and are now protected from editing. (See IRC log of approval discussion.)

See also: IRC log


Meeting (guest): OWL Working Group Teleconference [Scribe assist by trackbot-ng (guest)]
Date (guest): 13 February 2008 [Scribe assist by trackbot-ng (guest)]

Agenda amendments

Approve minutes

Ian Horrocks: can we accept previous previous minutes?

Peter Patel-Schneider: not really, but I'm not going to object
Rinke Hoekstra: don't think anything has changed in the minutes
Peter Patel-Schneider: actually, the minutes are just fixed

PROPOSED: accept previous previous minutes

Sandro Hawke: http://www.w3.org/2007/OWL/wiki/Teleconference.2008.01.30/Minutes
James Hendler: +1
Uli Sattler: +1
Rinke Hoekstra: ok
Michael Smith: +1 to accept, they improved much in the interim week
Bernardo Cuenca Grau: +1
Zhe Wu: +1
Rinke Hoekstra: +1

RESOLVED: accept minutes of http://www.w3.org/2007/OWL/wiki/Teleconference.2008.01.30/Minutes

Michael Smith: for previous minutes (Feb 6), new actions were not linked to tracker
Jeremy Carroll: -0
Peter Patel-Schneider: huh? my action was set up OK
Rinke Hoekstra: some links to issues are missing

PROPOSED accept minutes of http://www.w3.org/2007/OWL/wiki/Teleconference.2008.02.06/Minutes

Rinke Hoekstra: not a big deal, but we rejected minutes in the past on this ground
Peter Patel-Schneider: 30 Jan minutes were fixed up in the last hour or so
Ivan Herman: +1
Uli Sattler: +1
Rinke Hoekstra: (this is about feb 6 right?)
Boris Motik: +1
Alan Ruttenberg: +1 to close with carsten adding links
Bernardo Cuenca Grau: +1
Zhe Wu: +1
James Hendler: +1 (permanent plus one to accepting all minutes at any time)
Michael Smith: +1 to trusting Carsten
Carsten Lutz: sure
Jeremy Carroll: 0

RESOLVED: accept minutes of http://www.w3.org/2007/OWL/wiki/Teleconference.2008.02.06/Minutes

Alan Ruttenberg: Agenda amendment: Next monday meeting (1.5 weeks) should it be UFDTF or Imports?

Washington F2F

Peter Patel-Schneider: F2F page is at http://www.w3.org/2007/OWL/wiki/F2F2

Peter Patel-Schneider: F2F2 is right after OWLEd, registration is open

Peter Patel-Schneider: OWLED registration at http://www.webont.org/owled/2008dc/registration.html

Peter Patel-Schneider: OWLEd papers are due this friday

Alan Ruttenberg: +1 observers to be allowed

Ian Horrocks: so far, nobody has objected to observers, provided they stick to conditions in email

Bijan Parsia: ++1 to letting observers

Evan Wallace: are there restrictions on the number of observers?

James Hendler: +1 to chairs having to give permission to observers (so something is in the log)

Peter Patel-Schneider: reports that Clark&Parsia will sponsor dinner, NIST will possibly sponsor the room, but we still need fees

Joanne Luciano: wonders whether her organisation can sponsor (?)

Ivan Hermann: has never been to a W3C F2F with fees

Sandro Hawke: agrees that fees are very rare

Peter Patel-Schneider: suggests to drop lunches which will then reduce fees to about 50 dollars

Peter Patel-Schneider: with lunch fees would be about $100

James Hendler: I will have to see if my organization will let us pay a fee for this.
Alan Ruttenberg: breaks,even without food....

Bijan Parsia: F2F registration page is closed

Alan Ruttenberg: suggests to bring own foot/take food payment out of general bill

James Hendler: the Wiki page does not seem to have a link to registration - am I missing something?

Sandro Hawke: easy food is more valuable due to time saving

Peter Patel-Schneider: please sign up

Peter Patel-Schneider: sign up at http://www.w3.org/2007/OWL/wiki/F2F2_People

Action Item Status

Action 78:

Ian Horrocks: Action 78 seems to be done

Ian Horrocks: Action 80 seems to be done

Ian Horrocks: Action 81 seems to be done

Ian Horrocks: Action 82 seems to be done

Ian Horrocks: Action 83 seems to be done

Ian Horrocks: Action 84 seems to be done

Due and overdue Actions

Alan Ruttenberg: regarding Action 43 I will ...?

Alan Ruttenberg: +1
Alan Ruttenberg: 43 stays open, tracking Sandro. 62 closed.
Bijan Parsia: I can add stuff

Sandro Hawke: in 2-3 weeks, some code will be cleaned up, and people who want to put up test cases now, please talk to me

Ian Horrocks: Action 72

Alan Ruttenberg: will be continued, will change deadline

Proposals to Resolve Issues (15 mins)

Issue 3

Ian Horrocks: the proposal is to keep the status quo from OWL 1.0

Peter Patel-Schneider: q+ to agree to counterproposal, and thus postpone

Boris Motik: there is a counterproposal and we should postpone the discussion to next week

Boris Motik: use blank nodes in OWL11, so to be more compatible with RDF, use SPARQL-like semantics

Michael Schneider: +1 to boris

Peter Patel-Schneider: agrees with Boris to postpone to next week

Zakim: pfps, you wanted to agree to counterproposal, and thus postpone


Issue 91

Ian Horrocks: Boris has added ontology annotations, sent an email, ...

Ian Horrocks: proposes to close Issue ??

Alan Ruttenberg: sees open questions/possible issues regarding 91


Michael Schneider: q+ to postpone 91 to next week: want to think about alan's mail

Jeremy Carroll: mentions as example owl:imports which is not an annotation

Alan Ruttenberg: NOTE: The ontology-import construct owl:imports and the ontology-versioning constructs owl:priorVersion, owl:backwardCompatibleWith and owl:incompatibleWith are defined in the OWL vocabulary as instances of the OWL built-in class owl:OntologyProperty. Instances of owl:OntologyProperty must have the class owl:Ontology as their domain and range. It is permitted to define other instances of...
Alan Ruttenberg: ...owl:OntologyProperty. In OWL DL for ontology properties the same constraints hold as those specified for annotation properties in Sec. 7.1.
Alan Ruttenberg: http://www.w3.org/TR/owl-ref/#Annotations

Ian Horrocks: this sounds to complicated to be resolved today, we will postpone

Bijan Parsia: directive:= 'Annotation(' ontologyPropertyID ontologyID ')'
Bijan Parsia: | 'Annotation(' annotationPropertyID URIreference ')'
Michael Smith: q+ to suggest table be non-normative

ACTION: jeremy to send an email proposing a resolution of 91


ACTION: jeremy to send proposal for issue-91 ontology property


Issue 95

Ian Horrocks: there is now a table relating facets to datatypes

Zakim: msmith, you wanted to suggest table be non-normative
Peter Patel-Schneider: +1 to resolution

Michael Smith: would prefer this table being non-normative

Boris Motik: we need to restrict the datarange inside datatype restrictions

Bijan Parsia: +1 to boris

Boris Motik: without such a restriction, this table is meaningless

Michael Schneider: q+ to table
Alan Ruttenberg: alan doesn't recall that.
Alan Ruttenberg: I suggest resolving as is
Evan Wallace: Boris' suggestion sounds overly restrictive
Bijan Parsia: +1 to resolving it for now...we can revisit it as we consider, e.g., n-ary

Michael Schneider: is missing explanation of meaning in this table

Boris Motik: agrees

Boris Motik: -1 to resolve as is
Michael Schneider: -0
Jeremy Carroll: -0
Alan Ruttenberg: I vote with boris
Evan Wallace: -1
Uli Sattler: +1 to postpone with n-ary

Ian Horrocks: is disappointed :(

General Discussions (30 min)

Ian Horrocks: http://www.w3.org/2007/OWL/wiki/Teleconference.2008.01.30/Discussion

Fragments and Conformance

Ian Horrocks: do we want to retain OWL Lite

Bijan Parsia: we shouldn't retain OWL lite

Peter Patel-Schneider: +1 to not retaining OWL Lite

Bijan Parsia: OWL lite is difficult for an implementor

Alan Ruttenberg: +1 to not retaining OWL Lite (as currently defined) - perhaps retain the name and brand some other fragment
Boris Motik: +1 to bijan

Bijan Parsia: doesn't know of any OWL lite tools

James Hendler: thought he could suggest to let OWL lite go, but has heard people screaming out about this

James Hendler: sees external critique for dropping

Joanne Luciano: suggests to asks for input by OWL lite community

Alan Ruttenberg: q+ to suggest note explaining old OWL Lite's relation to OWL DL, emphasizing similar actual complexity

Ian Horrocks: this would be helpful

Bijan Parsia: I'll note that you can stay in owl lite
Bijan Parsia: even if we don't sanction it

Peter Patel-Schneider: if users were upset, they should speak out

Zakim: alanr, you wanted to suggest note explaining old OWL Lite's relation to OWL DL, emphasizing similar actual complexity
Bijan Parsia: +1

Alan Ruttenberg: hears that people using OWL lite are worried. Suggests to draft note with an explanation of how OWL lite relates to OWL DL, and suggest to not use it

Jeremy Carroll: q+ to suggest non computational advantages

...but to use OWL DL instead because it is so close and not computationally much cheaper

Bernardo Cuenca Grau: wonders what OWL lite1.1 would be and who would specify it

Bijan Parsia: Yeah, it's not like we can forbid staying inside owl lit!

Ivan Herman: wonders what the question is: do we want to say that OWL lite does no longer exist

James Hendler: we are not the entire OWL community, but we are being told explicitly that backward compatibilty is important

Alan Ruttenberg: I think we leave it exactly as is

James Hendler: we need to ask openly for input

Ian Horrocks: Jim, what does it mean to keep OWL lite?

Alan Ruttenberg: q+ to answer: Keeping it means no change.
Alan Ruttenberg: q+ to elaborate it *will* be compatible, as OWL 1.1 will be back compatible.

James Hendler: OWL lite is subset of OWL1.0, and we could update the this subset description???

Zakim: jeremy, you wanted to suggest non computational advantages
Alan Ruttenberg: q+ to also remind: Using OWL Light == Using OWL DL

Jeremy Carroll: sees 3 ways: (1) explicit deprecation, (2) silence (don't mention "lite", (3) Jim's suggestion

James Hendler: OWL lite 1.0 != OWL DL 1.1
Alan Ruttenberg: q+ to say "broken" isn't right word. "not useful to distinguish from OWL DL"
Sandro Hawke: (option 4 = apply the "OWL Lite" brand to some new fragment of OWL 1.1)

Jeremy Carroll: whereas OWL lite *is* close to OWL DL, there might be other advantages, eg, outside the implementors/complexity area

Jeremy Carroll: eg, it is easier to learn since it has fewer constructs

Michael Smith: q+ to ask what about owl lite would *need* updating
James Hendler: I note that the restriction of 1/0 on cardinality is one a number of developers have told me saves them a lot of time/effort because cardinality reasoning in general is difficult (in the open worlld)
Bijan Parsia: I'll note that my objections to owl lite is to what it does to users as well...in my experience what it encourages is *unfortuneate* from a user pov

Jeremy Carroll: Lite might also discourage the use of potentially oppressive constructs

Ian Horrocks: another option is to keep the name, but change its definition

Carsten Lutz: suggests to first discuss fragments in general before discussing them individually

Alan Ruttenberg: q+ to specify: any current OWL Lite document is a valid OWL 1.1 document


Carsten Lutz: we should have an idea about our opinion regarding fragments before putting loads of work in

Zakim: alanr, you wanted to answer: Keeping it means no change. and to elaborate it *will* be compatible, as OWL 1.1 will be back compatible. and to also remind: Using OWL Light == Using

Ian Horrocks: suggests to cut OWL lite discussion and move on to next point

Alan Ruttenberg: suggests to keep OWL lite to not force people/users to change what they are doing

Alan Ruttenberg: we keep a note saying that every OWL lite ontology is an OWL11 ontology

Alan Ruttenberg: suggests to not use "broken" for Lite but "not well motivated"

Carsten Lutz: +1 to alan
James Hendler: -1 to Alan
Michael Smith: +1 to alanr
Bernardo Cuenca Grau: +1
Evan Wallace: 0

Alan Ruttenberg: suggest to users to how to do things in the future, *if* they want to do things differently

Boris Motik: +1
Uli Sattler: +1 to alanr
Rinke Hoekstra: +1


Ivan Herman: ack m_schnei

Michael Schneider: OWL lite would be different to OWL lite11 due to punning

Jeremy Carroll: very god point!
Jeremy Carroll: ^o^oo
Alan Ruttenberg: I don't see that.

James Hendler: we made some changes to OWL DL which might cause problems with OWL lite

Ian Horrocks: I don't see it either
Michael Smith: can we action a description of such constructs?
Alan Ruttenberg: +1 to review and check that OWL Lite is still subset.
Alan Ruttenberg: Due diligence. I think we have consensus

James Hendler: we changed the semantics, so we should have change lite's semantics

Carsten Lutz: I don't understand. Why is OWL 1.0 Light without punning not a fragment of OWL 1.1, which *allows*, but doesn't *enforce* punning?

Ian Horrocks: we have true compatibility with OWL DL, and therefor also with OWL Lite

Alan Ruttenberg: suggests we do check Ian's compatibility understanding

Jeremy Carroll: +1 to jim
Michael Schneider: OWL-1.0-Lite onts would exactly mean the same in 1.1, *but* having OWL-1.1-Lite without punning would be strange

James Hendler: wants to see how this compatibility looks like before agreeing

Alan Ruttenberg: suggests to have a proposal to check in the future

Ian Horrocks: we need to record an action on somebody to carry out this check

Carsten Lutz: I understood Alan that there would be no OWL-1.1-Lite (only OWL-1.0-Lite mentioned in the 1.1 docs)

ACTION: alanr to write an email to propose such a check

trackbot-ng: Created Action 87 - Write an email to propose such a check [on Alan Ruttenberg - due 2008-02-20].

ACTION: jeremy to respond to punning and owl lite point

trackbot-ng: Created Action 88 - Respond to punning and owl lite point [on Jeremy Carroll - due 2008-02-20].

Ian Horrocks: over to "number of fragments"

James Hendler: the more fragments in a normative spec the worse it is:

James Hendler: more fragments causes confusion, so the bulk should happen in an informative and not in a rec track

Jeremy Carroll: q+ to suggest mechanisms ....

Ian Horrocks: is this a suggestion for all fragments

Alan Ruttenberg: q+ to ask whether others agree that not putting this in REC should be fine.

James Hendler: hm, perhaps Lite is different, and oracle might have a different opinion

Peter Patel-Schneider: q+ to say that we could handle OWL Lite with a pointer in the REC to a note
Zakim: jeremy, you wanted to suggest mechanisms ....

Uli Sattler: suggests to discuss properties of fragments first, then their number

Zakim: alanr, you wanted to ask whether others agree that not putting this in REC should be fine.
Jeremy Carroll: hp would probably support oracle if oracle wanted owl prime rec tracked; but i would need to check [Scribe assist by Jeremy Carroll]

Alan Ruttenberg: if we can agree that the fragments document is a note rather than REC track

Alan Ruttenberg: +1
Evan Wallace: Jeremy suggested that defining a standard way of specifying fragments could be REC tracked
Michael Smith: -1 to non-rec
Peter Patel-Schneider: +inf
Ivan Herman: -1
Evan Wallace: +1
Carsten Lutz: cannot answer this question

Ian Horrocks: who would agree fragments not being rec track?

James Hendler: +1 non rec track (except Lite)


Alan Ruttenberg: would like to know the positive benefit of making some fragment rec

Ivan Herman: even if fragments are only a note, some fragments could be rec track

Michael Schneider: I cannot answer before discussing with FZI first


Michael Smith: I think normative conformance statements for fragment is a benefity
Jeremy Carroll: +1 to zhe

Zhe Wu: oracle would like to see 1 fragment on the REC track and not just some kind of a note

Peter Patel-Schneider: q+ to ask which fragment?

Zhe Wu: this doesn't have to be OWL prime, but something sufficiently expressive and rule implementable

Carsten Lutz: which is the definition of OWL Prime

Zhe Wu: it could be bigger than OWL prime

Zhe Wu: but it couldn't be OWL Full

James Hendler: doesn't want to fight over *every* fragment

Alan Ruttenberg: why?

James Hendler: if we put a lot of them into REC track, then this will cause confusion

Alan Ruttenberg: ok

Ian Horrocks: looks at his watch

Ivan Herman: knows of other DB vendors who would like to support REC track fragment of OWL

Ivan Herman: Why aren't they in the WG? [Scribe assist by Alan Ruttenberg]
James Hendler: +1 to ivan - I have heard same from web app developers as well as vendors
Zakim: pfps, you wanted to ask which fragment?
Carsten Lutz: I will jump up and down :-)

Peter Patel-Schneider: if others want their fragment on REC track, then I want mine/other want theirs as well...

Alan Ruttenberg: q+ to address jumping up and down comments
Sandro Hawke: q+ to propose a market-driven selection mechanism

Jeremy Carroll: supports OWL Prime being REC track in case there are multiple vendors

Carsten Lutz: We don't know *which* fragment the other vendors want?!

Ian Horrocks: is this limited to *vendors* or can other people have a say?

Zakim: alanr, you wanted to address jumping up and down comments

Jeremy Carroll: in the past, the OWL DL people did such jumping and got their REC track

Carsten Lutz: I can repeat my argumentation for EL++
Michael Smith: q+ to respond to alanr
Carsten Lutz: (see early emails)

Alan Ruttenberg: would like to hear an argument for *why* a certain fragment should be in REC track

Alan Ruttenberg: then let Carsten make it
Michael Smith: exactly!
Alan Ruttenberg: We need to hear these
Carsten Lutz: Alan Ruttenberg: I made that already, but am happy to repeat it
Alan Ruttenberg: Maybe, if they are convincing
Alan Ruttenberg: Let's see
Michael Smith: to the extent that we all have to convince
Zakim: sandro, you wanted to propose a market-driven selection mechanism

Sandro Hawke: would like to see something like "for a fragment to be REC track, it needs at least 2 independent, public, complete implementations"

Bernardo Cuenca Grau: How do we measure the completeness of OWL Prime, for instance?

Ian Horrocks: how do we specify what an implementation for a fragment is?

Sandro Hawke: it should be serious, but doesn't need to be commercial

Alan Ruttenberg: Implementations should have advertised complexity
Carsten Lutz: exactly, ian
Bernardo Cuenca Grau: what if there is no advertised complexity?

Ian Horrocks: it could still be that quite a number would satisfy this

Zhe Wu: BigOWLIM + ORACLE for pD*
Zhe Wu: both are solid and scalable
James Hendler: Do you agree that 2 implementations is a good justification [Scribe assist by Alan Ruttenberg]

Sandro Hawke: plus "2 members of the WG needed per fragment, only 1 vote per member"

Alan Ruttenberg: the "implementation exists" seems to be a good criteria

Ivan Herman: pD* and more or less a level of OWL Prime is more or less implemented by Franz Inc and Ontotext, too
Zhe Wu: second that
Alan Ruttenberg: ok

James Hendler: disagrees with "implementation exists" is a good criterium because he would also like to see "vendors" and "companies" (??)


Alan Ruttenberg: company or open source would be a good compromise
Alan Ruttenberg: ?
Alan Ruttenberg: open source means that a company can take it up

Ian Horrocks: many different criteria, but Jim doesn't seem to like them

Michael Smith: jim, I am a corporate vendor. We've got DL-Lite and EL++ specific implementations that are being used by our clients

Sandro Hawke: asks Jim why many fragments are harmful?

Michael Smith: You have dl-lite? I want it !!!! [Scribe assist by Alan Ruttenberg]

Ian Horrocks: especially if it were clear that they are all subsets of OWL11 for which there also exists implementation

Carsten Lutz: I am sure I can also get commercial support for EL+, e.g. from SNOMED
Michael Smith: yes, to alanr

Alan Ruttenberg: tries to clarify the "what is a serious implementation"

James Hendler: wants to see oracle doing something being treated differently from what researchers are doing

James Hendler: REC track for the companies, notes for the researchers

Carsten Lutz: Sounds like OWL lite
Ian Horrocks: :-)
Alan Ruttenberg: ok. done now.

James Hendler: we need to be more careful with REC track than with notes, and we can't correct, eg, errors

Alan Ruttenberg: it's a pretty high bar to say two open source, independent implementations...

Ian Horrocks: but our fragments are not "new research project outcomes"

Alan Ruttenberg: I don't see that for el++, for instance, currently

James Hendler: we should learn from OWL

Carsten Lutz: Another example: DL-Lite was developed in cooperation with IBM, as far as I know.
Zhe Wu: product?

James Hendler: if all 7 fragments have the same status as EL+, then perhaps they can all go to REC track

Ian Horrocks: tries again to clarify the "what is a serious implementation"

Alan Ruttenberg: suggests to move discussion to email

Jeremy Carroll: the QA group adviced recently to have "as few parts as possible", and offers to check it out

ACTION: Jeremy to investigate QA group advised recently to have "as few parts as possible",

trackbot-ng: Created Action 89 - Investigate QA group advised recently to have \"as few parts as possible\", [on Jeremy Carroll - due 2008-02-20].

Other Business

Alan Ruttenberg: next week no Monday meeting because of holiday in US

Alan Ruttenberg: asks for suggestions for the week thereafter

Michael Smith: alanr, stick to schedule for week after, ufdtf
James Hendler: want to chat some time? [Scribe assist by Alan Ruttenberg]



Regrets:

Jeff Pan, Elisa Kendall, Markus Kroetzsch, Ratnesh Sahay