Warning:
This wiki has been archived and is now read-only.
Teleconference.2008.05.07/Minutes
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
Contents
- Present
- Peter Patel-Schneider, Rinke Hoekstra, Markus Krötzsch, Ivan Herman, Uli Sattler, Ian Horrocks, Bernardo Cuenca Grau, Michael Smith, Boris Motik, Michael Schneider, Achille Fokoue, Alan Ruttenberg, Sandro Hawke, Bijan Parsia, Jeremy Carroll, Evan Wallace, Jie Bao
- Regrets
- Carsten Lutz (travelling), Elisa Kendall (conflict with Semantic Web Deployment F2F), Martin Dzbor (project review meeting), Jeff Z. Pan (travelling)
- Chair
- Alan Ruttenberg
- Scribe
- Rinke Hoekstra
(Scribe changed to Rinke Hoekstra)
Admin
Roll Call
Agenda Amendments
Alan Ruttenberg: any amendments?
Next Week's TC
Alan Ruttenberg: next week's TC cancelled, no chairs, DL workshop
Alan Ruttenberg: any objections?
F2F3
Alan Ruttenberg: F2F3 in Boston on july 28 and 29
Alan Ruttenberg: please put yourself on the list if you're planning to attend or not
Previous Minutes
Alan Ruttenberg: needed some cleanup, heard from peter
PROPOSED: accept previous minutes
RESOLVED: accept previous minutes
Alan Ruttenberg: As Bijan recently said, if you weren't there you're actually a very good reviewer of the minutes: should be comprehensible
Pending Review Actions
Action 131
Alan Ruttenberg: implement decisions from the F2F2 for RDF mapping in particular
Alan Ruttenberg: has obviously been done
Alan Ruttenberg: solicit some reviewers to see whether this has been done (implementers, and someone involved in OWL Full)
Alan Ruttenberg: Michael? would you be willing to review
Michael Schneider: well, hmm, ok, yes..
Alan Ruttenberg: have the potential to affect owl full
Sandro Hawke: would it be helpful to create a colour-coded diff
Michael Schneider: differences are quite big
Michael Schneider: will simply read it
Alan Ruttenberg: Achille are you willing to take this on?
Achille Fokoue: won't be able to do this in the next two weeks
Alan Ruttenberg: that's no problem
Alan Ruttenberg: do you want to do this, and if so before when would you be able to do this?
Achille Fokoue: maybe end of may?
Alan Ruttenberg: would be happy personally, if you're willing
Sandro Hawke: will action Achille and Michael
Sandro Hawke: all documents? or parts? due date?
ACTION: schneider to review the changes made as result of Action 131 due May 20
ACTION: achille to review the changes made as result of Action 131 due May 30
Action 133
Alan Ruttenberg: is actually related, and the review would include that action as well
Alan Ruttenberg: if anyone disagrees that these actions aren't done, speak up
Action 132 and Action 129
Alan Ruttenberg: you laid out the possible options, do you have any particular idea about this, bijan?
Bijan Parsia: I didn't think that would be part of the action
Alan Ruttenberg: what we should do at least is ask if people could respond to vent their ideas/opinions to the options listed by bijan
Bijan Parsia: didn't feel like iterating all examples, if someone feels like adding examples, please do!
Due and Overdue Actions
Action 42
Action 43
Alan Ruttenberg: any update about this from jeremy, bijan, sandro?
Sandro Hawke: no progress, willing to work on this, can't get it to the top of my queue
Alan Ruttenberg: if anyone has test cases, please add them to the wiki
Bijan Parsia: if anyone could point me to a preferred format for this
Action 136
Alan Ruttenberg: jeremy?
Jeremy Carroll: don't know how much time is needed to discuss this, for next week is within the RIF timescale
Alan Ruttenberg: no meeting next week
Jeremy Carroll: any chance to fit 5 minutes in today?
Alan Ruttenberg: I'll put it in as the first issue
Action 142
Alan Ruttenberg: taken over by proposals from peter
Alan Ruttenberg: let's close that
Action 145
Alan Ruttenberg: Jeremy?
Jeremy Carroll: forgot this
Action 146
Jeremy Carroll: working on this
Action 144
Jeremy Carroll: this one has slipped my mind, and I am unlikely to push this forward
Alan Ruttenberg: does the current proposal satisfy our need in this area
Action 143
Alan Ruttenberg: didn't get to that
Bijan Parsia: one question about this last question
Bijan Parsia: I thought I sent this email, I already have some text about the old species in the primer. I just put a pointer to it
Alan Ruttenberg: thought was that the particular wording that jeremy had was quite nice
Alan Ruttenberg: if you think you have covered it, communicate this to jeremy
Alan Ruttenberg: I'll put an editorial note to put in the text that he had, so you can see whether you feel that adequately covers it
Review of RIF Compatibility
Jeremy Carroll: email linked from yesterday, mentioned a very few points. One point I forgot to put in the email
Jeremy Carroll: they use a generalised graphs ..something ..something ..bnodes ..literals
Jeremy Carroll: an RDF graph a subject is a bnode or uri, a predicate which is a uri, an obect which is a uri or literal
Jeremy Carroll: in their design they allow all three types in all three positions. This is a generalisation and quite an improvement, if you ask me
Alan Ruttenberg: any impact on serialisation?
Jeremy Carroll: it works with RDF graphs, it might mean that you can have a conclusion in RIF that can't be serialised
<! --Peter Patel-Schneider: can we do this too?
Jeremy Carroll: allright by me
Peter Patel-Schneider: surprised that alan isn't jumping up and down and screaming
Peter Patel-Schneider: destroys serialisability of everything
Jeremy Carroll: at the end of the email, they have text about OWL2 that could be more neutral
Jeremy Carroll: about punning
Jeremy Carroll: I would suggest that this WG should make that comment, it's not for me to say that by myself
Jeremy Carroll: the minimal review is that comment, along with some text like (...)
Jeremy Carroll: a very minor point, is that they haven't decided what sorts of entailments to include for RDF
Jeremy Carroll: simple entailment, rdf entailment, rdfs entailment
Jeremy Carroll: should say, don't bother thinking about RDF entailments
Jeremy Carroll: bulk of my comment is about a very silly thing actually... syntax is not standard
Jeremy Carroll: we might want to have some minor supportive text from the WG
Peter Patel-Schneider: comment about the inscrutable syntax choices has been pointed out to them several times
Peter Patel-Schneider: without much success
Bijan Parsia: I understand this to be part of the presentation syntax
Sandro Hawke: yes
Bijan Parsia: since it doesn't hit the wire, I don't care too much
Bijan Parsia: it's unclear whether our WG should care too much, unless we want to synchronise our spec. styles
Ivan Herman: it may affect one point. If we want to harmonise on the profile, we will be forced to take over that syntax in our description and pay the price
Sandro Hawke: I strongly disagree
Sandro Hawke: there's no grammar, you are not allowed to parse this syntax, it just helps to explain the semantics
Alan Ruttenberg: AS was parsed in OWL 1
Sandro Hawke: WG said you shouldn't
Alan Ruttenberg: is his specified as such?
Sandro Hawke: yes
(.. something about the way in which RIF deals with internationalised strings, rif:iri)
Jeremy Carroll: Sandro was arguing against including a comment on this topic (deviation from norms in presentation syntax)
Sandro Hawke: agnostic
Jeremy Carroll: many people have raised this, and it hasn't been taken notice of does suggest that it should be taken up as a WG issue
Jeremy Carroll: each WG has a task to take notice of other WG's
Bijan Parsia: just to go back to ivan's point. I agree that it is not to be serialised. We do have an interest, it is generally good to have the specs harmonised: some harmony is beneficial to reader
Bijan Parsia: it's still not a WG issue, jeremy is free to raise a last call issue
Bijan Parsia: we should focus on things that really impact our work
Bijan Parsia: not on just 'icky' stuff
Alan Ruttenberg: there's no show stoppers here. 1) don't waste your time on rdf 2) presentation syntax isn't standard
Alan Ruttenberg: we could send a note saying that we think you have done a good job etc. etc.
Alan Ruttenberg: reading both specs shouldn't be confusing, it would help to have a common syntax for readability reasons
Alan Ruttenberg: just show our interest on this issue, but no requirement
Peter Patel-Schneider: the syntax is not just in the presentation but also in the RIF-BLD
Sandro Hawke: I'm not sure what to make of that
Alan Ruttenberg: it's beyond presentation syntax
Sandro Hawke: I don't know what the concern is in RIF-BLD
Sandro Hawke: don't know if there's a problem with rif:iri
Sandro Hawke: my understanding is that RIF does not use IRIs as symbols (As owl and rdf). Instead it has a data mapping to go from IRIs to the arbitrary resources they stand for
Sandro Hawke: esp. michael kiefer preferred to do it like this
Alan Ruttenberg: do you think that's something we should be commenting on?
Peter Patel-Schneider: that's a good question... if we wanna fight, sure... but expect to fight
Alan Ruttenberg: my proposal is that we don't wan to fight, but say very clearly what we feel, and go on the record. Without saying that they *have* to fix the issue in the way we propose
Alan Ruttenberg: if peter doesn't mind writing up the note (removing jeremy's irritation etc.)
Sandro Hawke: one other comment, if you can say where it's actually harmful that would be good. If you want to have them change it, you should be clear on how much you would want this WG (owl) to slow down
Jeremy Carroll: it's not about RIF and OWL but about the specs that are already out there!
Alan Ruttenberg: How about a strawpoll, action a couple of people simply to write up some documentation in a neutral tone about what we saw and what we thought
Jeremy Carroll: we'll go on to the straw poll
STRAWPOLL
Rinke Hoekstra: +1
Alan Ruttenberg: neutral and positive mix...
Alan Ruttenberg: keep this action open?
Jeremy Carroll: yes
Ivan Herman: we did not officially approve jeremy's last point as a comment to the RIF group and the text they use regarding owl 2
Alan Ruttenberg: my idea was that the action would address this, and we would have some text that we could approve
Jeremy Carroll: perhaps approve on a draft via email, and send this draft before the deadline, vote on this post hoc, on the next telecon
Issues
Alan Ruttenberg: 15 minutes max
Alan Ruttenberg: on issues
Issue 85
Alan Ruttenberg: proposed to close as postponed, better use a better annotation syntax
Alan Ruttenberg: Alan Rector, who is the champion on this, was fine to postpone
Alan Ruttenberg: any questions abbout Issue 85
PROPOSED: to resolve Issue 85 as per http://www.w3.org/mid/61CBB11D-607F-40E0-AA1B-620C48E7E587%2540comlab.ox.ac.uk
RESOLVED: to resolve Issue 85 as per http://www.w3.org/mid/61CBB11D-607F-40E0-AA1B-620C48E7E587%2540comlab.ox.ac.uk
Issue 97
Alan Ruttenberg: question of whether or not the actual XSLT transformation is needed to be there, or whether the GRDDL could simply point to the mapping
Alan Ruttenberg: trick that I proposed does not actually work, as GRDDL does require an XSLT
Bijan Parsia: If I look at the GRDLL document it does not specify that you have to have an XSLT, it just mentions that you should have a transformation
Bijan Parsia: I would just like to have some textual support for your claim
Ivan Herman: bijan is right in terms of the recommendation. In fact, the GRDDL spec does not require the XSLT. However, as far as I know, no GRDDL implementation at the moment can handle anything else but XSLT...
Alan Ruttenberg: one of the objections to doing this was that we would have two normative rdf mappings. What we thought we could do is to assign an action to someone who would be happy to create an XSLT, and only publish it as a note of the wg
Alan Ruttenberg: would avoid any confusion about the status, and be friendly to anyone who would like to use that technology
Bijan Parsia: bad idea that the WG does implementation (especially as there are competing implementations such as the OWL API)
Bijan Parsia: best practice is to include it in your own software
Ian Horrocks: I find bijan's arguments quite persuasive on this. If it's not actually part of the GRDDL spec, I'm not sure why we're doing it
Ian Horrocks: I'm not quite sure what would be the note... algorithm? transformation?
Peter Patel-Schneider: there is a competing specification of the transformation ... the one we're writing
Jeremy Carroll: I wanted to take issue with bijan on the web retrievable issue. If you do object to this, you should have made an objection to the GRDDL spec. As it's actually a recommendation, there is a reason to take note of this
Jeremy Carroll: we can rely on the W3C of things not going away
Jeremy Carroll: my proposal would be that we could have two links, one to the actual spec (normative) and the xslt which is not normative (with a note on the top)
Alan Ruttenberg: chair hat off
(Ian Horrocks takes over as chair)
Alan Ruttenberg: I relate my understanding of what the point of this is
Alan Ruttenberg: same understanding as Jeremy's.
Alan Ruttenberg: the intention is that the XSLT is published, cached and then used to actually transform stuff to rdf/xml from xml. The spirit of this is that we put an XSLT transform there
Alan Ruttenberg: do not think it's damaging, do not think it should be blocked
Alan Ruttenberg: one of my objections to OWL/XML was resolved by adopting GRDDL
Alan Ruttenberg: and if we're not staying in the spirit of this, then I question whether we want the OWL/XML syntax
Alan Ruttenberg: continue next week?
Bijan Parsia: I don't see anything in the GRDDL spec that says that you have to retrieve something from the web
Bijan Parsia: I see the value of a web-retrievable transformation. We are not in that circumstance where we need that
Bijan Parsia: do people objecting to OWL/XML prefer to get some transformation somewhere from the web? I don't think this is a starter
Ian Horrocks: defer this until next week
Sandro Hawke: quick show of hands if anyone seconds Bijan's perspective?
Sandro Hawke: Should we have a strawpoll about retrievable but non-normative XSLT?
Ian Horrocks: is this about publishing a note, or is the note a uri that points to it, or describes it
PROPOSED: JJC's proposal for non-normative
Jeremy Carroll: in the OWL/RDF there's a bit of code that points to a GRDDL transform
Alan Ruttenberg: don't have time for this (chair hat on)
General Discussion: Imports
Alan Ruttenberg: Peter's updated proposal, Boris' comments on this
Alan Ruttenberg: didn't grab this from the web
Peter Patel-Schneider: latest proposal is to publish by location, do versioning by publishing in multiple spots. Implement this by writing this in section 3 of the syntax document
Boris Motik: the idea is to somehow split the imports from the actual locations where the ontologies are published
Boris Motik: question is, where is an ontology actually located?
Boris Motik: an ontology can have an ontology uri, and optionally a versioning uri. If it has any of these uris it should be published at a location that is equal to either one of these uri's
Boris Motik: imports points to a particular location, this location can be either equal to the ontology uri or the version uri that you want to import
Boris Motik: this procedure can be overridden for the purposes of caching
Alan Ruttenberg: any questions from anybody?
Alan Ruttenberg: have you thought about forward moving, if we decide to have something more involved as regards version information, does this preclude that?
Boris Motik: no, don't think so. You can actually encode additional information in the uri
Boris Motik: this is completely orthogonal.. you could abstract the whole thing by saying that you need some way of comparing two version uri's.
Boris Motik: you could encode numerical information and do comparison etc.
Uli Sattler: I was wondering in a similar direction
Uli Sattler: this mechanism would also allow me to always retrieve the latest version?
Boris Motik: the latest version is always at the location of the ontology uri
Boris Motik: when you create a next version, this current version goes somewhere else, and the new version gets put at the location
Boris Motik: if you want to import the latest version, you just point to the ontology uri.
Jeremy Carroll: this is very simple to spec, excellent, strong support
Alan Ruttenberg: it is still my intention to write a note offering this more complicated thing that shows that the simple mechanism doesn't handle this. Could we keep an issue open explaining use cases that I have, just to say that there's still an issue here
Alan Ruttenberg: easy to get out of sync in the OBO
Alan Ruttenberg: currently no way to repair that
Sandro Hawke: have you thought of mechanisms where you would have double version mechanisms, i.e. latest in 4.x latest in 5.x
Sandro Hawke: main production releases, beta releases, major / minor releases (latest of some obsolete version etc.)
Boris Motik: multiple ontology uri's, multiple default locations... this could be added, but in the existing proposal this is not captured
Peter Patel-Schneider: a previous version allows for multiple version uris, which I think would allow multiple branching, slightly more complex... don't know whether it's worthwhile allowing this
Peter Patel-Schneider: the WG decides things, and then people give in or object. What is this thing about having a minority report?
Jeremy Carroll: I'm pretty sure that sandro's use case is covered by this. I'm happy to take up an action to describe multiple versioning using this scheme
Sandro Hawke: to recast what I think Alan was wanting to do, was say: let's go ahead with something like this, but have some text in the spec or issues list that explains to people who wants something they need, that we don't provide. This can be consensus text
Peter Patel-Schneider: I thought I heard something about a separate note about this particular issue
Alan Ruttenberg: what I was saying was that having something more stronger is not something we have consensus about, but we could have something in a note that describes a more elaborate scheme
Alan Ruttenberg: didn't think this was controversial
Alan Ruttenberg: as sandro said, we could have some text about this, that could be taken up
Boris Motik: add a section about this, something similar to the 'oh, you could override the location in some way'
Boris Motik: gives people an idea on how to use this versioning. We could easily capture what should or could be added... what tools might want to do with this
Boris Motik: once we see what this looks like, it might be easier to comment on this. Unless anyone really objects, we could put this into the spec, and see how people feel about this
Jeremy Carroll: OWL2 is an improvement on OWL1 and that's the basic idea. OWL2 imports+versioning is an improvement on OWL1, but OWL3 will (hopefully) be an improvement on OWL2
Alan Ruttenberg: seems that this proposal is as far as the normative spec goes
Alan Ruttenberg: what I'm suggesting is that there's some work that has been done about use cases.. would be nice to have a record of this
Alan Ruttenberg: like what boris is saying.
Alan Ruttenberg: if I have time for a note, then we could discuss this at a later point
Michael Schneider: why have a normative part about this? Why not define the imports just as the imports closure
Michael Schneider: and just leave out the files stuff
Boris Motik: we actually started from that position. The member submission said exactly that... quite a few people objected. Are we prepared to backpaddle?
Alan Ruttenberg: strawpoll about this? General feeling about this proposal is that it's a positive step forward
STAWPOLL: are people comfortable having Boris put in the changes that he suggested?
Alan Ruttenberg: strong support for doing this
Alan Ruttenberg: put action on boris, ready to close
ACTION: bmotik2 to Implement the imports proposal as described in http://lists.w3.org/Archives/Public/public-owl-wg/2008May/0025.html
Alan Ruttenberg: thank you TF for putting effort on this!
Alan Ruttenberg: aob?
Boris Motik: Will need to defer this action to next week, because of the workshop
Alan Ruttenberg: UFDTF expect to have a telecon on monday
Alan Ruttenberg: adjourn