Chatlog 2008-06-18

From OWL
Revision as of 18:35, 18 June 2008 by Peter Patel-Schneider (Talk | contribs)

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

See original RRSAgent log and preview nicely formatted version.

Please justify/explain all edits to this page, in your "edit summary" text.

00:00:00 <pfps> PRESENT: bijan, m_schnei, Achille, bmotik, IanH, msmith, Zhe, uli, jie, Alan Ruttenberg, bcuencagrau, sandro, pfps, MartinD
16:48:42 <RRSAgent> logging to http://www.w3.org/2008/06/18-owl-irc
16:49:07 <pfps> pfps has changed the topic to: http://www.w3.org/2007/OWL/wiki/Teleconference.2008.06.18/Agenda
16:49:28 <pfps> ScribeNick: pfps
17:02:24 <pfps> Topic: Admin
17:02:55 <pfps> Subtopic: Roll call
17:05:31 <pfps> Subtopic: Agenda amendments
17:05:47 <pfps> alanr: move imports from resolve to discuss
17:05:57 <pfps> alanr: quick update on testing
17:06:16 <pfps> Subtopic: Approve Minutes
17:06:31 <pfps> PROPOSED: Accept Previous Previous Minutes (04 June)
17:07:18 <pfps> pfps: formatting looks different
17:07:29 <pfps> alanr: other opinions?
17:07:50 <pfps> uli: 4th are OK, 11th worse
17:08:19 <uli> no
17:08:21 <pfps> alanr: review minutes next week for formatting issues
17:08:28 <pfps> alanr: objections? - none
17:08:58 <pfps> Subtopic: F2F3
17:09:07 <pfps> alanr: please indicate registration status
17:09:20 <pfps> Subtopic: Action item status
17:09:20 <pfps> ACTION-160
17:09:51 <pfps> alanr: ACTION-160 pending
17:10:07 <pfps> boris: everything OK except status of top in EL++
17:10:13 <pfps> alanr: Zhe?
17:10:31 <uli> ...say againb
17:10:36 <pfps> Zhe: OK with current status
17:10:51 <pfps> alanr: should bottom role be added separately where OK?
17:10:59 <msmith> q+ to ask about dl-lite
17:11:07 <pfps> boris: they should be added as pairs as they are converses
17:11:25 <pfps> msmith: is top OK in dl-lite
17:11:57 <pfps> boris: might change complexity, but I don't know, so an investigation would be needed to add it
17:12:08 <pfps> alanr: add an issue for this?
17:12:34 <msmith> +1 to close action. this can be discussed on list.
17:12:35 <pfps> uli: action on me to ask Diego whether dl-lite can accept top
17:12:55 <pfps> alanr: ACTION-160 closed
17:13:07 <pfps> ACTION-42
17:13:14 <pfps> alanr: Bijan not here so pass
17:13:18 <pfps> ACTION-150
17:13:23 <alanr> http://www.w3.org/2007/OWL/wiki/InternalizedString
17:13:39 <pfps> alanr: has anyone looked at the document
17:13:55 <pfps> alanr: intent of action is to also start discussion
17:14:14 <pfps> jie: I gathered information
17:14:39 <pfps> alanr: can you also initiate discussion with RIF WG?
17:14:46 <pfps> jie: OK
17:14:58 <pfps> boris: is this all going to a new document?
17:15:17 <pfps> jie: the page is not part of the spec, but it will lead to something
17:15:40 <msmith> Yes, I thought it was to be a spec, in case we beat RIF forward
17:15:43 <pfps> alanr: Ivan was suggesting a mini-specification, but the WG stance was not clear
17:16:32 <pfps> alanr: there should be some publication in some form
17:17:05 <pfps> alanr: there would then be something for both WGs to point to
17:17:19 <pfps> ACTION-155
17:17:42 <pfps> alanr: Ivan not here so postpone
17:17:52 <pfps> ACTION-156 and ACTION-157
17:18:06 <pfps> alanr: I'm behind so postpone
17:18:15 <pfps> Topic: Issues
17:18:25 <pfps> Subtopic: ISSUE-109
17:18:48 <pfps> alanr: what namespace to use for owl xml syntax
17:19:02 <pfps> alanr: Ivan's email summarizes the situation
17:19:11 <pfps> alanr: vote on the issue
17:19:36 <bijan> Hi
17:20:35 <pfps> PROPOSED: Use one namespace (vote 1) two namespaces (vote 2) or don't care (vote 0)
17:20:44 <pfps> pfps: 1
17:20:45 <bmotik> 2
17:20:46 <jie> 0 
17:20:49 <uli> 1
17:20:49 <IanH> 1
17:20:49 <bijan> 1
17:20:53 <bcuencagrau> 0
17:20:55 <m_schnei> 2 (fzi)
17:21:05 <MartinD> 2
17:21:14 <alanr> Ivan - 2
17:21:38 <msmith> vote 0 (C&P)
17:21:49 <alanr> 2 (science commons)
17:21:51 <pfps> Achille: if we use the same namespace, will there be different things using the same URI
17:22:29 <pfps> bijan: no overlap (URI vs Qname), no overlap even discounting this
17:22:39 <Zhe> 0 (Oracle)
17:22:39 <Achille> 1
17:22:44 <jie> 0(RPI)
17:22:46 <Achille> 1 (Achille)
17:22:50 <Achille> 1 (IBM)
17:22:52 <msmith> revised vote 1 (C&P)
17:23:29 <bijan> yes
17:23:29 <uli> yes
17:23:40 <uli> 1 (Manchester)
17:24:09 <MartinD> revision ... 2 (Open University)
17:24:12 <bmotik> Oxford - 0
17:25:14 <msmith> I see 4x0 3x1 4x2
17:25:23 <bijan> 4 x 1
17:25:49 <alanr> 3x0 4x1 4x2
17:25:50 <alanr> ?
17:27:48 <pfps> alanr: and the result is a tie!  the chairs will figure out if it really is a tie
17:28:16 <pfps> Subtopic:  ISSUE-112
17:28:26 <pfps> alanr: names of top and bottom properties
17:28:40 <pfps> boris: topObjectProperty, bottomObjectProperty, ...
17:28:56 <pfps> boris: ..., topDataProperty, bottomDataProperty
17:29:07 <bmotik> They are called owl:TopObjectProperty, owl:BottomObjectProperty, owl:TopDataProperty, owl:BottomDataProperty
17:29:39 <pfps> PROPOSED: Resolve ISSUE-112 adding top and bottom properties with the names owl:TopObjectProperty, owl:BottomObjectProperty, owl:TopDataProperty, owl:BottomDataProperty
17:29:41 <bijan> Nope
17:30:05 <m_schnei> if there is bottomDataProperty, why isn't there owl:DataNothing?
17:30:07 <pfps> alanr: Does this mess up the "single" root of property hierarchies?
17:30:30 <bmotik> +1
17:30:32 <bijan> +1
17:30:32 <msmith> +1 to resolve issue-112 as stated
17:30:33 <jie> +1
17:30:36 <alanr> 0
17:30:36 <bcuencagrau> +1
17:30:42 <m_schnei> +1
17:30:47 <uli> +1
17:30:49 <MartinD> +1
17:30:51 <pfps> pfps:+1 to resolve issue-112 as stated
17:30:52 <Zhe> +1
17:30:54 <Achille> 0
17:30:56 <IanH> +1
17:31:07 <pfps> RESOLVED: Resolve ISSUE-112 adding top and bottom properties with the names owl:TopObjectProperty, owl:BottomObjectProperty, owl:TopDataProperty, owl:BottomDataProperty
17:32:08 <pfps> m_schnei: is there a difference between Nothing and the empty datatype?
17:32:27 <pfps> m_schnei: do we need a name for the empty datatype?
17:33:04 <pfps> boris: for properties we need the two bottom properties to allow for syntactic typing of expressions
17:33:31 <pfps> m_schneid: what about empty datatype name?
17:33:43 <alanr> complementOf(rdfs:literal)?
17:33:49 <pfps> boris: we can have it - but there is already syntax for it
17:34:04 <pfps> bijan: i go either way
17:34:32 <pfps> Subtopic: ISSUE-21 and ISSUE-24
17:35:54 <pfps> IanH: there has been extensive discussion leading to a "resolution"
17:36:21 <pfps> IanH: see http://lists.w3.org/Archives/Public/public-owl-wg/2008Jun/0126.html
17:37:23 <pfps> pfps: wording has problems for me - also incompatibleWith gives syntactic invalidity
17:37:39 <pfps> boris: this has been there a few weeks
17:38:29 <pfps> pfps: my problem was the wording of Alan's message - syntactic invalidity
17:38:41 <pfps> boris: incompatibleWith has been in a while
17:39:18 <pfps> boris: still a should - only change is to call the problem a syntactic invalidity
17:40:30 <pfps> alanr: two cases - 1 - imports of two different versions of the ontology - 2 - importing incompatibleWith ontologies
17:40:51 <pfps> alanr: were treated differently - changed to be the same way
17:41:24 <pfps> alanr: syntactic invalidity was borrowed from case 1
17:42:00 <pfps> boris: intention was to treat both cases the same
17:42:09 <IanH> no! ??
17:42:38 <pfps> boris: now they are both treated the same way 
17:43:18 <pfps> pfps: I don't consider this to be a syntactic problem - instead it is something else
17:44:02 <pfps> IanH: then what should happen
17:44:28 <pfps> boris: should allows exceptions
17:44:45 <pfps> IanH: both cases should be the same
17:44:55 <pfps> pfps: both are the same
17:45:28 <pfps> pfps: syntactic validity *should* not involved a should
17:45:41 <pfps> IanH: let's take this to email
17:45:57 <m_schnei> in reverse RDF mapping, what happens if incompatibleWith occurs? is rdf graph rejected to be syntactically invalid OWL DL?
17:46:38 <pfps> alanr: status of incompatibleWith on third-party ontologies
17:46:43 <pfps> boris: resolve together
17:47:09 <pfps> IanH: let's push for a single solution
17:47:25 <pfps> Subtopic: ISSUE-108
17:47:37 <pfps> alanr: what to name the profiles
17:47:53 <alanr> http://lists.w3.org/Archives/Public/public-owl-wg/2008May/0120.html
17:48:01 <pfps> alanr: contenders - one- and two-letter suffixes (see http://lists.w3.org/Archives/Public/public-owl-wg/2008May/0120.html)
17:48:47 <pfps> alanr: [lists the possibilities]
17:48:50 <bijan> I'll note that I was misremembering earlier :( I just brain farted about e.g., owl:ObjectProprety. I'll send an email :(
17:48:52 <bmotik> DDL is bad because it sounds too much like "distributed description logics"
17:49:08 <pfps> alanr: I prefer the one-letter versions
17:49:31 <uli> ..the link above seems to be broken
17:49:41 <IanH> Not for me!
17:50:07 <uli> ...the ")" was the culprit, it works
17:50:42 <alanr> 1) OWL2E 2) OWL2 E 3) OWL E
17:50:58 <bmotik> In general, we have write "OWL 2" (with a space)
17:51:03 <bmotik> s/write/written
17:51:14 <alanr> 4) OWL 2 E
17:51:17 <bmotik> Yes
17:51:18 <pfps> pfps: yes
17:51:21 <bijan> OWL 2e
17:51:29 <m_schnei> OWL E 2
17:51:37 <alanr> 5) OWL 2E
17:51:43 <m_schnei> I mean this seriously...
17:51:51 <alanr> 6) OWL E 2
17:51:55 <bijan> But isnt' the first version of OWL E?
17:51:59 <IanH> OWL E2?
17:52:20 <IanH> OWLE2?
17:52:23 <alanr> 2 be or not 2 be
17:52:32 <IanH> Not serious
17:52:35 <bijan> I just meant that I don't understand 6)
17:53:13 <alanr> 1) "OWL2E" 2) "OWL2 E" 3) "OWL E" 4) "OWL 2 E" 5) "OWL 2E"  6) "OWL E 2"
17:53:55 <uli> ...no it wouldn't help us
17:54:00 <pfps> m_schnei: so consider OWL 2 Lite, ...
17:54:05 <bijan> "OWL E 2" sounds like the second version of OWL E, not the E version of OWL 2 :)
17:54:14 <bmotik> 4
17:54:18 <m_schnei> 3
17:54:19 <pfps> Straw Poll: put in preference (single vote)
17:54:21 <bcuencagrau> 4
17:54:25 <alanr> 3
17:54:27 <Zhe> 3 or 4
17:54:28 <pfps> pfps: 4
17:54:33 <bijan> 5 (lowercase)
17:54:36 <uli> 1 or 4
17:54:41 <IanH> 4
17:54:51 <MartinD> 4 or 3
17:54:52 <bijan> OOo, OWL^2^e
17:54:55 <msmith> 2
17:55:01 <jie> 1
17:55:06 <Achille> 3
17:55:06 <bcuencagrau> hey, what about the single vote?
17:55:33 <m_schnei> ok, I am going to withdraw 6 ;-)
17:55:55 <pfps> msmith: we are talking about spacing
17:55:56 <uli> good idea!
17:56:54 <IanH> We didn't consider hyphens!
17:56:55 <alanr> 1) E, D, R 2) EL, DB, LP for  EL++, DL Lite, OWL R
17:57:24 <uli> there was an alternative for the 2letter version for LP
17:57:52 <bmotik> 2
17:57:54 <bcuencagrau> 2
17:57:55 <pfps> alanr: if the choice is two-letters we will later decide on which letters
17:57:59 <Zhe> 1
17:58:00 <alanr> 1
17:58:01 <MartinD> 2 chars
17:58:04 <m_schnei> 1
17:58:09 <Achille> 2
17:58:10 <pfps> Staw Poll: one letter vs two letters
17:58:11 <pfps> pfps: 0
17:58:12 <msmith> 1 to reduce search space
17:58:12 <uli> 1
17:58:15 <bijan> 0
17:58:19 <IanH> 2
17:58:21 <jie> 1
17:58:48 <bijan> I guess I could change to 1
17:59:07 <bmotik> +q
17:59:09 <IanH> I should own up to the fact that 3 twos came from Oxford
17:59:12 <pfps> alanr: and the result is ............. 1 letter ...... by a hair
17:59:23 <bcuencagrau> +q
17:59:44 <alanr> phone crapped out. back in a sec
17:59:55 <Achille> +1 for boris
18:00:09 <pfps> boris: we should consider which one-letter names
18:01:11 <pfps> boris: OWL 2 E could be OWL 2 T (for tboxes)
18:01:19 <bmotik> OWL 2 Tax
18:01:24 <uli> eee
18:01:54 <pfps> bernardo: E doesn't mean anything, we need a better letter for it
18:02:01 <IanH> Agree with Bernardo!
18:02:45 <m_schnei> expand "tbox"... "T" for defining "[T]erminologies
18:02:47 <Zhe> second Bijan
18:03:01 <uli> ...a bit of mnemonic helps, but there can be too much in a name!
18:03:02 <bmotik> +q
18:03:08 <pfps> bijan: the names should just be "names", i.e., with no inherent meaning
18:03:14 <bcuencagrau> right
18:03:17 <pfps> boris: then use A,B,C
18:03:19 <pfps> bijan: OK
18:03:28 <IanH> Disagree with A,B,C!
18:03:44 <pfps> uli: a little bit of mnemonic is useful
18:03:49 <IanH> +1
18:04:22 <bijan> I wouldn't select A, B, C, obviously. THe current names are better!
18:04:35 <m_schnei> A=[A]ssertionbox, B = data[B]ase, C = [C]lasses
18:04:41 <pfps> alanr: think about this during the week and a proposal may show up for next week
18:04:44 <IanH> For EL and R, I don't see what is wrong with EL and R
18:05:01 <pfps> Topic: General Discussion
18:05:15 <pfps> List of normative datatypes
18:05:30 <alanr> http://www.w3.org/2007/OWL/tracker/issues/126
18:06:04 <pfps> boris: the current list of datatypes includes lots of XML schema datatypes, some of which are problematics
18:06:12 <pfps> boris: Jeremy had a paper on this
18:06:22 <MartinD> MartinD has left #OWL
18:06:43 <pfps> boris: decimal datatype not useful - no division - rational better
18:07:15 <pfps> boris: float and double are even worse - finite but huge, with bad mathematical properties
18:07:34 <pfps> boris: these could end up with strange consequences
18:07:35 <alanr> q+ to ask whether if we don't use these in class expressions, could we leave them for annotations
18:07:49 <pfps> boris: Jeremy points out that operations are not associative on these 
18:08:03 <pfps> boris: let's use instead rational and/or real
18:08:31 <pfps> boris: date datatypes have some problems
18:09:09 <pfps> boris: pattern facet can be applied to numbers with hard-to-predict results
18:09:31 <pfps> boris: these datatypes are in XML schema to handle input and don't match reasoning
18:09:42 <pfps> uli: I wildly agree with Boris
18:09:50 <pfps> uli: RacerPro people also agree
18:10:13 <pfps> alanr: what about using them in annotations?
18:10:52 <pfps> msmith: several different problems, should we split them?
18:11:03 <pfps> msmith: rational/real is already on the table
18:11:05 <msmith> msmith: I suggest breaking this into multiple issues, for reals we have ISSUE-87 with proposed resolution http://www.w3.org/2007/OWL/wiki/OWL_Rational
18:11:50 <pfps> pfps: input could have odd datatypes, provided they could be mapped to "better" datatypes
18:12:17 <alanr> q+ to ask if approximating annotation values would mean you wouldn't get back what you put in sometimes.
18:12:33 <pfps> boris: numerics issues intertwined - if rational and real are included then this solves some of the problems
18:12:56 <m_schnei> so what would then be the mandatory types? string, integer, rational, real? something else? (reasoning about quaternions in 3d apps might be cool :))
18:12:58 <pfps> boris: against allowing input-only datatype
18:13:51 <pfps> alanr: couldn't annotation values just be left alone?
18:14:08 <pfps> alanr: wouldn't approximating input values lead to roundtripping problems
18:14:38 <pfps> alanr: what about OWL accepting experimental data (which will be in some existing format)
18:17:08 <pfps> boris: input data should be mapped into better datatypes prior to input to OWL
18:17:36 <alanr> link http://www.comlab.ox.ac.uk/people/boris.motik/publications/mh08datatypes.pdf is broken
18:17:57 <msmith> use http://www.comlab.ox.ac.uk/people/boris.motik/pubs/mh08datatypes.pdf
18:17:53 <pfps> boris: could also allow input that looks like "2.5"^^xsd:float, but this would be translated into the real / rational 2.5
18:18:12 <pfps> pfps: that was essentially my proposal
18:18:59 <pfps> bijan: can't give up input with double/float/... but ...
18:19:34 <pfps> boris: in most cases nothing bad would happen, but double/float are finite and having a datarange of float is dangerous
18:19:44 <pfps> boris: for dates even the names are problematic
18:19:56 <pfps> bijan: this breaks existing ontologies
18:20:07 <pfps> boris: but OWL 1 only had integer and string
18:20:11 <msmith> one doc says sting & integer, one says everything but duration
18:20:18 <pfps> bijan: it allowed others and they were used
18:20:30 <pfps> alanr: more investigation needed 
18:20:53 <pfps> alanr: my main concern is scientific use of OWL
18:21:31 <pfps> uli: for numerics - internal space can be larger (input) but input space can be as in XML
18:21:50 <pfps> uli: could even have xsd:float being internally the same as real
18:22:10 <msmith> not xsd:float though, really xsd:decimal
18:22:15 <pfps> alanr: what about rounding
18:22:34 <IanH> cerebra
18:22:47 <pfps> boris: network inference datatype reasoner did this remapping
18:23:08 <IanH> s/network inference/cerebra/
18:23:21 <pfps> boris: we could just do the remapping internally
18:23:23 <alanr> q+ to ask if we could have a small writeup of this proposal that we could shop around to interested parties
18:23:34 <msmith> q+ to react to remapping
18:23:42 <m_schnei> computer languages generally have LongFloat libraries, with arbitrary length decimals
18:23:43 <pfps> boris: rounding would not be a problem (if nothing is done on the value)
18:24:21 <pfps> alanr: could we have a writeup or a pointer on the mapping solution
18:24:39 <pfps> boris: I'll do it
18:25:58 <pfps> Topic: Other Business
18:26:03 <pfps> Subtopic: Testing
18:26:09 <msmith> msmith: I've started by using the WebOnt tests.  See http://www.w3.org/2007/OWL/wiki/User:MikeSmith#WebOnt
18:26:22 <msmith> ... rather than move them all, I moved one of each test type and am beginning to develop the supporting infrastructure around that
18:26:30 <msmith> ... there are a few different hierarchies for tests (e.g., type, "species", etc)  See http://www.w3.org/2007/OWL/wiki/Category:Test_Case
18:26:55 <msmith> ... people can base new tests on what's there, but it shouldn't be considered stable yet and you're probably better waiting
18:27:21 <pfps> alanr: questions?
18:27:38 <pfps> alanr: Adjourn