From OWL
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

Jeremy Carroll: JeremyCarroll has changed the topic to:

Accept previous minutes

Peter Patel-Schneider: minutes are not great

PROPOSED: accept previous previous minutes

Michael Smith: there was a lot of zakim-ness, but they were sufficient for me to understand what I missed.
Peter Patel-Schneider: potentially out-of-order stuff
Peter Patel-Schneider: also strange blue right parenthesis

Alan Ruttenberg: Jeff, need more work?

Jeff Pan: tried to incorporate Peter's comments

Peter Patel-Schneider: around Issue 124 start

Alan Ruttenberg: consider it not ready

PROPOSED: Thank Jeremy Carroll for his exemplary service to the WG and wish him well in his new employment

Alan Ruttenberg: Jeremy's last meeting. we all thank him!

Michael Schneider: +1
Peter Patel-Schneider: +2
Rinke Hoekstra: +3
Alan Ruttenberg: +1
Jeff Pan: +1
Ivan Herman: +infinite
Boris Motik: +1
Michael Smith: +1
Ian Horrocks: +1
Uli Sattler: +1
Elisa Kendall: +1
Evan Wallace: +!
Zhe Wu: +1
Peter Patel-Schneider: no - you can vote for yourself

RESOLVED: Thanks Jeremy Carroll for his exemplary service to the WG and wish him well in his new employment

Action items status

Alan Ruttenberg: pending review actions

Action 143 Put editorial note in profiles document

Action 42 postponed

Michael Smith: that's fine, I already emailed sandro about it

Action 43 Develop scripts to extract test cases from wiki. closed.

Action 139 Sheperd/coordinate the patching process (per Issue 119)

Ian Horrocks: good progress made. don't mind leaving it open

Alan Ruttenberg: estimation?

Ian Horrocks: sometime before next F2F

Michael Schneider: The action itself can be closed. Issue 119 should be left open. expect to have the first draft somewhere in June so we have enough time before F2F. I am working on it.


Issue 21 and Issue 24 Imports and Versioning

Alan Ruttenberg: I want IanH to chair this portion

Ian Horrocks: already have text based on Peter, Boris, AlanR's discussion

Ian Horrocks: alanr has some issues

Alan Ruttenberg:
Alan Ruttenberg:

Alan Ruttenberg: first one, not importing multi version of the same ontology. second, owl:incompatibleWith

Alan Ruttenberg: owl:incompatibleWith
Boris Motik: Alan, can you please repeat the first point?

Ian Horrocks: not clear to me that we can resolve it now

Alan Ruttenberg: first point in 0176.html

Michael Schneider: Issue 21 about import, it is not clear to me

Rinke Hoekstra: my questions were adequately answered by Boris' answer
Rinke Hoekstra: to my email

Boris Motik: answer to alanr's comment. if two onotlogies are marked incompatible, it is better to say nothing when multi version imported. current spec says nothing when multi version imported. if you need validation, it is out of the scope.

Jeremy Carroll: +1 to emphasising positives
Michael Schneider: Issue 24 is actually: "Make it be that importing two ontologies which are noted to be incompatible leads to an inconsistent ontology.
Alan Ruttenberg: ack bmotik

Ian Horrocks: all versions are treated as advisory, rather than formal

Boris Motik: you get the union of multi versions. SPEC provides no mechanism for detecting this. you can implement on top of OWL 2.

Alan Ruttenberg:
Ian Horrocks: ack alanr

Boris Motik: i implemented what I thought that we agreed.

Ian Horrocks: alanr, are you arguing about what you want, or the process

Alan Ruttenberg: at the workshop, we did not have a solution. But later peter sent a follow-up email

Alan Ruttenberg: An ontology SHOULD NOT import multiple versions of the same ontology,
Alan Ruttenberg: i.e., different ontology documents with the same ontology URI but that
Alan Ruttenberg: do not share an owl:versionInfo annotation value

Ian Horrocks: alanr, do you like the SPEC to include precise statement on what will happen if two versions of the same ontology are imported?

Jeremy Carroll: q+ to respond to SHOULD

Alan Ruttenberg: I like to say what peter said that ontology should not import multi versions

Boris Motik: Actually the SPEC is precise on that.

Peter Patel-Schneider: q+ to respond to inconsistent issue
Alan Ruttenberg: Although, I am not suggesting now necessarily that there is inconsistent. This would require more work.
Evan Wallace: are different versions of the same ontology implicitly incompatibleWith each other?
Michael Smith: For SHOULD NOT, see bullet 4.
Ian Horrocks: ack m_schnei

Jeremy Carroll: A bit disappointed that import TF is not decisive

Ian Horrocks: To be fair, everyone thought that we have agreed. and then implemented what we agreed.

Ian Horrocks: Jeremy said that SHOULD is the right thing to say

Uli Sattler:'s occupied

Peter Patel-Schneider: I agree with Jeremy

Peter Patel-Schneider: I am happy with the way it is. Put SHOULD in to make some folks in the WG happy. right now, SHOULD is not there

Uli Sattler: ahaa
Ivan Herman: :-)
Ian Horrocks: ack bmotik
Rinke Hoekstra: Should should make me happy

Boris Motik: sure. that is ok. if we can close the issue

Jeremy Carroll: for me the should is quite ugly too ...
Ian Horrocks: ack alanr
Alan Ruttenberg:

Ian Horrocks: think it is useful. maybe we can converge and resolve it

Alan Ruttenberg: think it is just editorial. owl:incompatibleWith, shall we discuss it as a separate issue?

Michael Schneider: currently, AFAIU, all ontology properties are treated as annotations, in particular owl:incompatibleWith

Ian Horrocks: it is semantic free

Alan Ruttenberg: it carries some weight on what people think their tools should do

Alan Ruttenberg:
Rinke Hoekstra: +1 to IanH on explaining the SHOULDs, MAYs and MUSTs

Ian Horrocks: like to have some text clarifying "SHOULD." At least add a pointer.

Uli Sattler: perhaps we can have a brief explanation that explain why this "should"
Uli Sattler: e.g., that different version could lead to inconsistencies
Michael Schneider: distinguishing between model-theoretic "reasoning" semantics, and "usage" semantics

Boris Motik: I changed the text. took out the offending paragraph. add "SHOULD NOT". . Hope it solves the problem

Jeremy Carroll: IanH raised a good point that SHOULD is advisory

Alan Ruttenberg: but why should we say anything about what happens when you don't do a SHOULD

Boris Motik: for SHOULD, MAY, ..., there is a disclaimer at the beginning

Ian Horrocks: ack bmotik
Alan Ruttenberg: no, no, I am not suggesting that! I'm against changing model theory for incompatibleWith
Alan Ruttenberg: yes!
Rinke Hoekstra: that would be great
Michael Schneider: i would also be against giving a model-theoretic semantics to owl:incompatibleWith
Peter Patel-Schneider: +1 to boris's suggestion
Ian Horrocks: ack alanr

Alan Ruttenberg: don't want to change semantics as well

Michael Schneider: (but in OWL Full, this property of course *has* a model-theoretic semantics ;-))

Boris Motik: prefer lower case and do a review. Later, change "should" systematically

Jeremy Carroll: I prefer that 'should' doesn't occur except as SHOULD
Rinke Hoekstra: +1 to JeremyCarroll
Evan Wallace: If we know we mean SHOULD now lets say it
Alan Ruttenberg: use
Alan Ruttenberg: ought
Alan Ruttenberg: s/should/ought/
Alan Ruttenberg: +1

Jeremy Carroll: you can always rephrase "should." It makes things simple for the readers

Ian Horrocks: we all in agreement now?

Alan Ruttenberg: Boris, are you going to put something similar in the document for incompatibleWith?

Peter Patel-Schneider: +1 to taking a vote :-)
Rinke Hoekstra: +1 to vote
Uli Sattler: ...and in this case?
Uli Sattler: me too!

Ian Horrocks: are we voting on should => SHOULD, or incompatibleWith?

Alan Ruttenberg: resolve SHOULD NOT import multiple ontologies or owl:IncompatibleWith
Rinke Hoekstra: I suggest we vote on the issues first
Jeremy Carroll: +1 to caps or something
Evan Wallace: +1 to caps

PROPOSED: spec should state that an ontology SHOULD NOT import two incompatible versions

Rinke Hoekstra: We have two official issues, let's vote on them separately.
Michael Schneider: +1 to SHOULD NOT
Rinke Hoekstra: This is to resolve Issue 24 right?
Boris Motik: Proposed text: Furthermore, O should not import an ontology O' with a version URI vu if O contains an ontology annotation owl:incompatibleWith with the value vu.
Jeremy Carroll: +0 to proposal

PROPOSED: SPEC should state that an ontology SHOULD NOT?

Ian Horrocks: getting into too much details on wording. it would be better if you guys figure this out precisely off line

Ivan Herman: +1 to Ian
Carsten Lutz: +1
Zhe Wu: +1
Uli Sattler: +1
Jeff Pan: +1 Ian
Alan Ruttenberg: +1
Michael Schneider: +1, this is a wording issue
Evan Wallace: +1 on narrowly wordsmithing this res offline
Jeff Pan: action?

Ian Horrocks: we had enough discussion on this issue. come back next week

Michael Schneider: do we want an action on this?

Issue 124

Issue 124 (newly open) The complement of a datarange is defined relative to the whole data domain

Alan Ruttenberg: consensus is this is how things are.

Alan Ruttenberg: ack m_schnei

Michael Schneider: Boris's comments are valid. The only thing is we could have this thing in the primer. suspect people will ask how to do complement on just the data type.

Michael Schneider:

PROPOSED: The complement of a datarange is defined relative to the whole data domain (close as resolved Issue 124)

Alan Ruttenberg: m_schnei can put a comment in the primer

Evan Wallace: Need to advise users about this somewhere.

Jeremy Carroll: for OWL2 FULL, complementOf should be on the data type.

Alan Ruttenberg: ack uli

Michael Schneider: in owl full, if you take complement of xsd:integer, then you get owl:Thing minus xsd:integer

Uli Sattler: this piece of advice perhaps is too detailed for primer. should go somewhere indeed

Michael Schneider: my example would be the same in OWL Full
Alan Ruttenberg: cookbook ;-)
Ivan Herman: +1000 to Uli
Ivan Herman: wiki
Alan Ruttenberg: ack bmotik
Jeremy Carroll: _:x rdf:type rdfs:Datatype
Jeremy Carroll: _:x owl:complementOf T(DR)
Alan Ruttenberg: ack JeremyCarroll
Jeremy Carroll: (I am leaving now ... bye)

Alan Ruttenberg: time is past. let us continue on email

Issue easy keys

Alan Ruttenberg: just to check we are on the same page on easy keys. both easy keys/top bottom added to spec, with formal addition to language based on vote. can we do a straw poll

Peter Patel-Schneider: don't think your description match minutes

Peter Patel-Schneider: the straw poll wording in the minutes does not mention documentation change

Peter Patel-Schneider: the straw poll in the minutes does not mention document changes at all
Michael Schneider: lets have to distinct polls
Rinke Hoekstra: That's Issue 112

Ian Horrocks: one question on top/bottom, do we agree on the name?

Alan Ruttenberg: not

Rinke Hoekstra: (names, I mean)
Alan Ruttenberg: ack bmotik

Alan Ruttenberg: add them as top and bottom, . and an editorial note saying that names are not final

Boris Motik: implementing universal role is hard. I am not convinced it is "easy." I like to keep it separate from easy keys

Alan Ruttenberg: where do we stand on easy keys?

Alan Ruttenberg: ack pfps. should we add easy key?

Peter Patel-Schneider: not aware of implementation of easy keys

Michael Schneider: missing major stakeholders, shall we defer the decision?

Peter Patel-Schneider: so I feel that the documents should mention that easy keys may be yanked if implementations are not produced
Alan Ruttenberg: ack ivan

Alan Ruttenberg: my sense is that majority of this WG are stakeholders and they are for it

Alan Ruttenberg: ack bmotik

Boris Motik: thinking about implementing easy keys. not trivial, should not be impossible. We should have larger scale evaluation though.

Peter Patel-Schneider: agree with Boris that implementation situation wrt easy keys is different from the vast bulk of OWL 2

Alan Ruttenberg: we should have general discussion on these next week

Issue 109

Issue 109: What is the namespace for elements and attributes in the XML serialization

Alan Ruttenberg: 1) namespace itself. 2) should we reuse the same namespace

Ian Horrocks: Sorry, but I have to leave now.

Ivan Herman: namespace in terms of XML, and namespace used in RDF/OWL are very different. I am in favor of two different namespaces to avoid problems for OWL/XML

Rinke Hoekstra: +1 to ivan
Boris Motik: +1
Alan Ruttenberg: 1.
Alan Ruttenberg: 2.
Alan Ruttenberg: 3.
Alan Ruttenberg: 4.
Michael Schneider: +1 to ivan: different things want different URIs
Peter Patel-Schneider: I don't see any problems with sharing, but I really don't care

Ivan Herman: we decided to use owl namespace for the whole thing. so 1) is ruled out. don't care other three

Rinke Hoekstra: prefer 'xml' to be in there

Alan Ruttenberg: suggest 3. The year there give us possibility to evolve)

Uli Sattler: I guess Bijan has, but he isn't here today
Rinke Hoekstra: but history shows that we don't evolve
Alan Ruttenberg: straw poll : how about for the OWL-XML namespace

Uli Sattler: just curious to hear what problems will come up if we only have one namespace

Ivan Herman: there are lots of discussion in XML world of what exactly the semantics of namespace is. a word of caution is not to mix up things

Alan Ruttenberg: ack uli
Rinke Hoekstra: sidenote: the owl namespace has a month in it as well

Uli Sattler: then it seems like something we should not decide now. need more information before we can make a decision.

Rinke Hoekstra: e.g. and currently in the syntax spec

Ivan Herman: why it is a big problem to separate the two?

Rinke Hoekstra: Ivan is right wrt the hash mark

Ivan Herman: if we decide to have a different one. I don't care which

Michael Schneider: people will click on the XML URI, and will expect to get to something related to the XML, not related to OWL in general
Alan Ruttenberg: straw poll : how about for the OWL-XML namespace
Uli Sattler: I think Bijan could be one
Uli Sattler: ok ok
Uli Sattler: 0
Alan Ruttenberg: 1
Zhe Wu: 0
Achille Fokoue: 0
Boris Motik: 0
Michael Schneider: +1 to have distinct URIs
Peter Patel-Schneider: 0
Michael Smith: +1
Ivan Herman: I have a slight preference for alt. 2
Jeff Pan: 0
Rinke Hoekstra: +0.5 the separate namespace is fine, but I think Sandro 'd like a month in there
Carsten Lutz: 0
Evan Wallace: +1 for URI 3 for now

Issue 112

Alan Ruttenberg: Issue 112 What name to give to Universal Property. Consensus: not trying meaningful name

Uli Sattler: but this is related to the discussion before and thus deferred?
Rinke Hoekstra: ... but what if we won't have a UP?
Uli Sattler: ...108?

Issue 104

Issue 104 disallowed vocabulary OWL 1.1 DL does not have a disallowed vocabulary

Michael Schneider: in old OWL SPEC, have disallowed vocabulary. However, in the new RDF mapping document, we don't have something similar. e.g. having rdf:List is allowed in the new spec but not in the old spec

Alan Ruttenberg: ack bmotik
Ivan Herman: must not ...
Alan Ruttenberg: MUST NOT
Alan Ruttenberg: ack m_schnei

Boris Motik: don't think this belong to the mapping document. Check section 2.2.of FS

Alan Ruttenberg: ack ivan

Ivan Herman: boris, fully agree. OWL/XML namespace should not have any new terms. it is irrelevant

Boris Motik: it does have elements from OWL/XMl schema. I will change it after tele conf

Michael Smith: on tests. make progress next week (before next F2F). I am willing to be aggregation point

Rinke Hoekstra: thanks