OWL Working Group Meeting Minutes, 19 December 2007

DRAFT. Currently Under Review

See also: IRC log


PROPOSED: Accept Previous Minutes

RESOLVED: Accept Previous Minutes

Pending Action Items

Boris Motik: They did
Boris Motik: Ian added them
Peter Patel-Schneider: yes, I'm waiting for Sandro to push the button
Jeff Pan: bijan, thx
Bijan Parsia: So this means no publication until after Jan 2 (/me notes)
Bijan Parsia: No I mean that the publication moratoriam; it ends on Jan 2

Alan Ruttenberg: Action 4, are we using redirects?

Sandro Hawke: we can cancel the action

Peter Patel-Schneider: there are *still* people who haven't signed on to the wiki

Ian Horrocks: there are still people without wiki accounts

Peter Patel-Schneider: just a comment
Bijan Parsia: How do people scribe if they don't have accounts?
Bijan Parsia: So that's a forcing function
Markus Krötzsch: A simple captcha can prevent spam. There are simple solutions to that.
Jeremy Carroll: (note steve battle is HP's alternate, and so far has not done anything, other than attend F2F because of HP's internal decisions about representation etc.)
Jeremy Carroll: (note steve does not have an account)
Sandro Hawke: MarkusK, captcha's don't prevent spam -- we get spam from real humans.
Markus Krötzsch: OK, human spam is of course different.

Alan Ruttenberg: update the deadline for Action 23

Alan Ruttenberg: mid-january for Action 23

Ian Horrocks: Is it just me or is Alan a very long way away?
Peter Patel-Schneider: I don't remember anything related to that issue at the F2F.
Jeremy Carroll: I think this action should continue

Alan Ruttenberg: do we want punning on object and datatype properties?

Bijan Parsia: does it matter?

Evan Wallace: jeremy received this action at the FTF

Alan Ruttenberg: describe the problem in detail

Ian Horrocks: There is no call next week!

Alan Ruttenberg: update the deadline for Action 48

Bijan Parsia: I recall asking for a grammar

Ian Horrocks: zhe will write OWLPrime spec in a similar format to others

Zhe Wu: trying to figure out if RDFS 3.0 vocabulary can be mapped to DL-Lite or EL++

Zhe Wu: summary could be in wiki in January

Zhe Wu: update deadline for Action 54

Peter Patel-Schneider: I don't want a meeting in any case
Bijan Parsia: Uli is out sick

Alan Ruttenberg: continue Uli's actions Action 55 and Action 56

Ian Horrocks: there were pending review items

Ian Horrocks: pending means people want confirmation that it is finished

Bijan Parsia: I was confused by that as there were extant imports proposals (e.g., from my email from before the f2f)

Alan Ruttenberg: accept Action 36, Action 40, Action 41 as closed

Bijan Parsia: OWL 1.1 uses constants to describe atomic values, such as strings or integers. These are encoded as in the Turtle specification[TURTLE].
Bijan Parsia: constant is the same as the literal production in the Turtle specification [TURTLE]

Alan Ruttenberg: can you put the revision of wiki page regarding Action 57

Alan Ruttenberg: there should be a note in the tracker about the action

Jeremy Carroll: For clairfication: this was about actions not issues?

Alan Ruttenberg: goal is for someone to understand what the action is and how it was resolved

Alan Ruttenberg: people can review easily

Alan Ruttenberg: pending review actions completed

Local Arrangements for Second F2F

Peter Patel-Schneider: put a link to wiki [1]

Peter Patel-Schneider: location is Gaithersburg, MD

Alan Ruttenberg: personally encourage people to attend OWLED


RDF Mapping Task Force

Alan Ruttenberg: establish RDF mapping task force

Ian Horrocks: members will be Jeremy, Peter, Jonathan, Ian, Alan, Boris

Alan Ruttenberg: task force will create a wiki page

Alan Ruttenberg: address open issues regarding RDF mapping

Alan Ruttenberg: ensure backward compatibility

Jeremy Carroll: is Jonathan a WG participant? [Scribe assist by Jeremy Carroll]
Alan Ruttenberg: he is in process of joining [Scribe assist by Jeremy Carroll]

Imports discussion

Alan Ruttenberg: let's discuss imports

Boris Motik: imports are by ontology URI but physical URI may be different

Bijan Parsia: BTW, I don't understand the comment "Bijan, I believe you were supposed to put this somewhere on the Web page, right?"

Boris Motik: propose to have oracles to map ontology URIs to physical URIs

Boris Motik: a file based oracle plus hints in the imports

Boris Motik: if all else fails try ontology URI

Bijan Parsia: XML catalog has similar mechanism

Bijan Parsia: Protege and TBC has similar mechanisms

Alan Ruttenberg: +1 for specification
Boris Motik: Can I answer this please?

Jeremy Carroll: do we need to specify the mechanism for interoperability?

Rinke Hoekstra: Topbraid uses the xml:base instead of an ontology uri to resolve imports to local files
Peter Patel-Schneider: I concur with Jeremy
Jeremy Carroll: why does boris's proprosal benefit interoperability? [Scribe assist by Jeremy Carroll]
Jeremy Carroll: all the tools have this sort of caching mechanism, of course. [Scribe assist by Jeremy Carroll]
Jeremy Carroll: but for interop we should specify the Web as where your documents are stored [Scribe assist by Jeremy Carroll]
Alan Ruttenberg: +1 to sandro - web über alles
Jeremy Carroll: and it is a tool matter how caching, temporary copies etc. etc. [Scribe assist by Jeremy Carroll]
Jeremy Carroll: -1 to sandro
Sandro Hawke: I'd be much more comfortable with this mechanism if it were sold as "here's a manual override" if you really need it. [Scribe assist by Sandro Hawke]

Sandro Hawke: we should first try ontology URI then manual override

Bijan Parsia: Does this include file:// uris?

Boris Motik: in practice, you don't retrieve ontologies from the web

Bijan Parsia: Does this mean you need to synch ontology name and location?
Sandro Hawke: I was NOT saying "web first then fallback". Manual override comes first.

Boris Motik: most of the time they are local

Alan Ruttenberg: yes, understood that, apologies for misphrasing
Jeff Pan: -1 on most of the time ontologies are local

Boris Motik: trying web first is fine but it is important to say there is another mechanism

Rinke Hoekstra: XML uses the xml:base to state where an xml file is supposedly located

Alan Ruttenberg: chair hat off, a use case for imports regarding versioning

Alan Ruttenberg: multiple versions of the same ontology

Sandro Hawke: jeremy (on irc): * jeremy is happy with Sandro's clarification [since /me things are not recorded]

Alan Ruttenberg: ontology is same but location for different versions change

Alan Ruttenberg: a differetn use case then caching

Rinke Hoekstra: I think that different versions are *different* ontologies

Alan Ruttenberg: local file does not work well with Web

Alan Ruttenberg: have the mapping in a Web file

Rinke Hoekstra: ok, I'll chew on this for a bit
Bijan Parsia: I tend to treat different versions, esp during development, as "the same" ontology
Alan Ruttenberg: treating different versions as different ontologies doesn't work for the users I've worked with on this. [Scribe assist by Sandro Hawke]
Boris Motik: true, but using different-location to solve the versioning problem isn't a great solution either. [Scribe assist by Sandro Hawke]

Boris Motik: if we want to solve versioning problem we need to look at it independent of imports

Bijan Parsia: name == location is only true of OWL DL; owl full doesn't require that

Peter Patel-Schneider: solution to versioning is same as W3C versioning

Peter Patel-Schneider: as I proposed in the F2F

Alan Ruttenberg: what is an ontology name good for?
Peter Patel-Schneider: see the imports page
Alan Ruttenberg: peter, could you q up to describe your proposal quickly

Bijan Parsia: I work with command line tools and local files

Bijan Parsia: putting a mapping on the Web does not work for those cases

Alan Ruttenberg: I'm reluctant on imports solution without considering versioning

Rinke Hoekstra: Addendum to Bijan's remark: confusion is even worse because there's nothing that relates the RDF representation of an ontology with the classes/axioms in that ontology other than that they are both 'in' the same file
Bijan Parsia: +1 to Rinke

Alan Ruttenberg: putting a map might not work always but there should be a possiblity to do so

Alan Ruttenberg: even if the contents might change ontology identity name remains same

Bijan Parsia: Addendum to me: having a tool fail an import because the ontology URI is distinct from the location really frustrated me (Protege 4 had this for a while)
Carsten Lutz: ian +1

Ian Horrocks: since alanr is discussing I'll chair this discussion

Jeremy Carroll: bijan, why does your use case require the spec to address it?

Bijan Parsia: It's relevant in that right now the discussion in the spec is confusing and tools vary a lot
Bijan Parsia: So anything we do to help the tools converge helps a lot
Alan Ruttenberg: the problem is that people use multiple tools, jeremy
Rinke Hoekstra: +1 to at least clear up the confusion and *then* decide on what we want to standardise or leave to the tools

Boris Motik: spec should be precise with regards to how imports work

Alan Ruttenberg: +1 to boris

Boris Motik: otherwise developers come up with different solutions

Bijan Parsia: I agree with boris, this is a spec issue

Jeremy Carroll: +1 to clarification

Alan Ruttenberg: we are not that far away from a solution

Alan Ruttenberg: get interested people discuss the issue

Alan Ruttenberg: after we have consensus that this is a spec issue

Alan Ruttenberg: maybe do a straw poll

STRAWPOLL (guest): The OWL 1.1 Spec should talk about hot to physically locate ontologies [Scribe assist by Sandro Hawke]
Bijan Parsia: +1 to speccing imports
Jeff Pan: +1 spec issue
Peter Patel-Schneider: +1 part of the spec
Rinke Hoekstra: +1 to spec
Evren Sirin: +1 to spec
Carsten Lutz: 0

Ian Horrocks: majority thinks this is a spec issue

Jeremy Carroll: (do we want to specify how to locate local copies?
Jeremy Carroll: (I perhaps could be persuaded by peter, since I thought I was agreeing with him)
Jeff Pan: (sorry, have to leave now - bye)
Alan Ruttenberg: bye jeff
Rinke Hoekstra: (have to leave as well...)

Jeremy Carroll: it might do harm to put in spec

Jeremy Carroll: better handled in tools

Jeremy Carroll: that tools should be allowed to compete on non-interoperability issue [Scribe assist by Jeremy Carroll]
Bijan Parsia: I gave an example how OWL 1.0's specing resulted in interop problems
Bijan Parsia: With test cases
Alan Ruttenberg: Multiple tools use was the other
Bijan Parsia: (interop problem IMHO)
Boris Motik: +1 to bijan
Sandro Hawke: I agree with Jeremy that it should be in the spec if and only if it's necessary for interoperability. So let's figure out whether it's necessary.... [Scribe assist by Sandro Hawke]

Ian Horrocks: people involved will dicuss importing issue via email

Ian Horrocks: action item for alanr, 3 weeks from now, to arrange imports discussion

Ian Horrocks: passing the chair hat back to alanr

ACTION: Alan to organize task force on imports -- due 3 weeks


Alan Ruttenberg: Let's go through the raised issues and decide if we want to accept or reject the issue

Alan Ruttenberg: Issue 9 by pfps

Peter Patel-Schneider: I have nothing to say on the issue.

Alan Ruttenberg: bernardo. dow we want to accept this issue?

Bernardo Cuenca Grau: I'm not sure if this is RDFS semantics or OWL-Full semantics

Bernardo Cuenca Grau: DL-Lite is a proper superset

Bernardo Cuenca Grau: worth responding to

Boris Motik: I believe this is an issue: under the Full semantics, RDFS is not embeddable into any DL fragment
Bijan Parsia: I don't understand why we're accepting that....what document is this an issue against?

Alan Ruttenberg: Accept Issue 9

Alan Ruttenberg: Issue 10

Evren Sirin: This is the exact same problem raised in Issue 22

Alan Ruttenberg: reject Issue 10 as duplicate of Issue 22

Carsten Lutz: +1 alan

Alan Ruttenberg: Issue 22

Alan Ruttenberg: I've come up with similar use casse

Alan Ruttenberg: at least two other people find it useful

Alan Ruttenberg: so we can put it in the functional syntax

Boris Motik: Isn't Issue 23 somehow subsumed by our general approach to blank nodes?
Jeremy Carroll: (we're still on 22 not 23)
Bijan Parsia: It seems crazy to me without reflection into all syntaxes

Ian Horrocks: by accepting an issue we are just saying it is not crazy to discuss it

Alan Ruttenberg: accept Issue 22

Alan Ruttenberg: moving to Issue 23

Jeremy Carroll: suggest merge with Issue 3

Alan Ruttenberg: Issue 3 could be slightly different

Alan Ruttenberg: is there anybody seconding to accept?

Alan Ruttenberg: reject Issue 23 on the basis there is noone seconding

Alan Ruttenberg: moving to Issue 24

Alan Ruttenberg: combining onotlogies that are incompatible with each other cause inconsitency

Alan Ruttenberg: owl:backwardcompatibleWith property

Evan Wallace: seconding to accept Issue 24

Alan Ruttenberg: Issue 24 accepted

Alan Ruttenberg: Issue 47

Bijan Parsia: at the face to face uli and I are tasked to work up a proposal on this front
Peter Patel-Schneider: not raised by me - instead taken from the Webont issues

Alan Ruttenberg: accept Issue 47 based on easy keys proposal

Alan Ruttenberg: second bijan to accept Issue 47

Alan Ruttenberg: accepted Issue 47

Bijan Parsia: seconded, move on

Alan Ruttenberg: moving to Issue 56

Alan Ruttenberg: Issue 56 accepted

Jeremy Carroll: a minute or two please to redial

Alan Ruttenberg: Issue 60 regarding wine ontology

Boris Motik: Isn't this taking the political correctness a tad too far?
Bijan Parsia: Isn't this something for the UFDTF
Peter Patel-Schneider: is there *any* universal example?
Bijan Parsia: In the primer proposal peter and I are working we're exploring alternative domains

Ivan Herman: we should accept

Jeremy Carroll: I second it

Evan Wallace: I think that the example was only used in Guide

Alan Ruttenberg: Issue 60 accepted


Alan Ruttenberg: let's adjourn if there is no other important issue

Zhe Wu: happy holidays!

Alan Ruttenberg: next meeting in two weeks

Last modified on 2 January 2008, at 10:44