See also: IRC log
<uli> aha, this explains things
<scribe> ScribeNick: msmith
<bmotik> > Zakim, who is on the phone?
<sandro> it's always slow at the top of the hour.
<bcuencagrau> Zakim ??P8 is me
<pfps> ack ??P5
<IanH> ack +0186527aacc
<IanH> ack ??P2
ianh: no agenda amendments
<pfps> 4 june minutes look acceptable
RESOLUTION: Accept Previous3 Minutes (04 June)
<pfps> 11 june minutes look acceptable
<IanH> +1
PROPOSED: Accept Previous2 Minutes (11 June)
<uli> +1
RESOLUTION: Accept Previous2 Minutes (11 June)
<pfps> 18 june minutes are *perfect* :-)
PROPOSED: Accept Previous Minutes (18 June)
<IanH> +1
RESOLUTION: Accept Previous Minutes (18 June)
ianh: clarify status on http://www.w3.org/2007/OWL/wiki/F2F3_People
ianh: on action-160 wasn't there
question on top/bottom in profiles? keys in profiles?
... there was an action on uli re: top/bottom in profiles
uli: I sent an email on top/bottom in dl-lite. diego?
calvanese: dl-lite has no top
concept... there is no point to having it. we don't believe it
would impact properties, but there is not point
... if it doesn't change computation properties, it is just by
chance
... you don't gain any expressivity
ianh: its already that it doesn't add expressive power to DL
calvanese: yes, b/c you have nominals, that might not apply to profile which is strict subset
uli: reason to add is not to add
expressivity, it is to add useful syntactic sugar
... e.g., rooting a property hierarchy from a top property
ianh: with profiles, ruling
things out is costly rather than having them
... we should only rule things out if e.g., they have adverse
impact on properties
msmith: +1 to ianh
calvanese: I partially agree. adding construct gives indication it is to be used. this may have bad impact, even if it can be simulated with existing constructs
<JeffP> +1 calvanese
calvanese: similar argument for dl-lite profile
bmotik: only profile now
including top/bottom is EL++
... I don't think property must be in profile for editor to
hang things off it in UI
<uli> 1-
ianh: we had discussion about
top/bottom being useful and addressed if it *tempts* users in a
negative way
... it seems we can have it in dl-lite
calvanese: I'd like to check the details on whether we can have it
ianh: revisit this in future
telecon
... top/bottom is in el++
bmotik: not in owl-r
ianh: should we action someone to investigate easy keys
bmotik: no. its clear no easy
keys in dl-lite
... I added it to owl-r
<scribe> ... unknown for EL++
jeffp: top/bottom in el++ ?
bmotik: yes, checked with Carsten
jeffp: it doesn't have nominals
ianh: yes, presumably it doesn't hurt
bmotik: yes, it doesn't hurt
jefffp: what about el+
<bcuencagrau> EL++ without nominals
bmotik: what's el+
jeffp: el+ is supported by CEL
<JeffP> ok
ianh: a bit off topic, we're only
concerned with EL++ profile, not other fragments
... interesting that CEL doesn't support all of EL++ since
we'll need to follow-up moving forward the recs
calvanese: follow-up on keys in dl-lite, and boris's comments on it adding recursion. we'd like to see some version of keys, could we consider a restricted version.
ianh: are you willing to take action
<scribe> ACTION: calvanese to investigate top/bottom roles in dl-lite [recorded in http://www.w3.org/2008/06/25-owl-minutes.html#action01]
<trackbot> Created ACTION-162 - Investigate top/bottom roles in dl-lite [on Diego Calvanese - due 2008-07-02].
ACTION: calvanese to investigate easy keys in dl-lite
ACCEPT ACTION-160 as completed
ianh: action-155
<pfps> could we have a pointer to the document from the ACTION-155 page?
ianh: there is a document, we
also need implementation
... yes, we should add pointer to doc to action
<ivan> no
ianh: bump date forward for
action-155 pending arrival of an implementation?
... ok, that's what we'll do
... action-156, action-157
alanr: push them both a week
ianh: ok
ianh: proposal to resolve says "per pfps email and subsequent discussion", are we really here? it doesn't seem complete
alanr: we're close, have 1 issue
open
... is inconsistent independent of header? bmotik and I
disagreed
... it may be case inconsistency is noticed by user, not
maintainer, we'd like to state this
bmotik: one ontology saying
something about another is recipe for disaster
... breaks encapsulation. let's people say anything about
anything
<alanr> how is this different from having axioms on a class in two different ontologies?
<Rinke> Not sure whether this has anything to do with the issues per se? Seems that the issues are being overloaded with side-issues that prevent them from being resolved.
bmotik: detecting these incompatibilities and maintenance could get out of hand
<alanr> detecting is trivial
alanr: I'm not persuaded
bmotik: allowing one ont to say something about another seems to me as a conceptual hack
<Rinke> +1 to separate issue!
alanr: you're arguing conceptual
integrity vs. use case from personal experience
... we can spin this off to another issue and resolve the
rest
uli: +1 on separate issue
... +1 to bmotik that this will open can of worms and may be
difficult to explain behavior
ianh: I see what you mean, just as you don't have control over another on, you may not have control over statements saying what onts are incompatible
bmotik: already what we have is an improvement
alanr: not sure that's the case for owl 1
bmotik: but there was no semantics
alanr: yes, problem was no teeth to semantics
bmotik: tool is more that welcome to do this. seems to be extrapolating from one use case
ianh: given we have agreement other than this, can we move forward closing ISSUE-21 and ISSUE-24 and open new issue to discuss versioning?
alanr: incompatible with, not versioning
<pfps> fine by me
<Rinke> +1
ianh: yes, incompatibleWith
<IanH> PROPOSED: resolve Issue 21 and Issue 24 Imports and Versioning, per update from Boris, Peter's email and subsequent discussion, modulo opening new issue on incompatibleWith
bmotik: if we move forward splitting, I think we should take everything out
<bmotik> -q
alanr: I disagree unless strong opposition. it would be a step backwards
ianh: if we resolve in favor of your approach, doesn't that mean ripping out what's there now?
alanr: ontology header is better than nothing, if we remove it we may have to readd it later
bmotik: I'd prefer to discuss if we need incompatibleWith at all
alanr: it seems we're now moving backwards
pfps: I suggest going as proposal says, discuss incompatible with as separate issue
bmotik: out of document?
pfps: minimal change to current doc. it is an interim state, even if no one likes it
bmotik: ok
<IanH> PROPOSED: resolve Issue 21 and Issue 24 Imports and Versioning, per update from Boris, Peter's email and subsequent discussion, but open new issue on status of incompatibleWith
<pfps> +1 to resolve this way
<bmotik> +1
<alanr> +1
<uli> +1
<Rinke> +1
<IanH> +1
<ivan> 0
+1
<baojie> 0
<Ratnesh> +1
<bcuencagrau> +1
<IanH> RESOLVED: resolve Issue 21 and Issue 24 Imports and Versioning, per update from Boris, Peter's email and subsequent discussion, but open new issue on status of incompatibleWith
<Zhe> +1
<alanr> happy happy
<alanr> joy joy
<bmotik> ACTION to bmotik2: Update the strucutral spec according to resolution of ISSUE 21 and ISSUE 24
<trackbot> Sorry, couldn't find user - to
<bmotik> ACTION: bmotik2 to Update the strucutral spec according to resolution of ISSUE 21 and ISSUE 24 [recorded in http://www.w3.org/2008/06/25-owl-minutes.html#action02]
<trackbot> Created ACTION-163 - Update the strucutral spec according to resolution of ISSUE 21 and ISSUE 24 [on Boris Motik - due 2008-07-02].
ianh: ISSUE-81 can be resolved
using bmotik's proposal to use an alternative vocabulary for
reification
... any reasons not to resolve?
<IanH> PROPOSED: resolve Issue 81 Reification of Negative Property Assertions, per Boris's email
<pfps> +1 to proceed apace
<bmotik> +1
<Rinke> +1
<IanH> +1
+1
<bcuencagrau> +1
<ivan> +1
<Ratnesh> +1
<uli> +1
<IanH> RESOLVED: resolve Issue 81 Reification of Negative Property Assertions, per Boris's email (http://lists.w3.org/Archives/Public/public-owl-wg/2008Jun/0156.html)
<alanr> +1
<ivan> happy happy
<Zhe> +1
<JeffP> +1
<Rinke> http://lists.w3.org/Archives/Public/public-owl-wg/2008Jun/0171.html
ianh: brief revisit of profile
naming (ISSUE-108)
... (as in Carsten's email) at least OWL-R and OWL-EL names are
ok, DL-Lite needs a name
... Carsten proposed calling it owl-db, but that's likely to be
contentious
<Zhe> :_
msmith: why can't we call it dl-lite?
<calvanese> unmute me
<calvanese> unmute me
<alanr> we want to market to a larger community!!
ianh: owl-lite is deprecated, owl dl-lite seems rather long winded
<sandro> "OWL2 Lite" ?
<alanr> OWL-D
calvanese: we believe name owl-db
would be suitable, since owl-r people like owl-r lets use
owl-db
... owl-d doesn't evoke anything related to dl-lite
<alanr> OWL-I
calvanese: I am not in favor of owl-d
zhe: is this profile specific for
db modeling integration and nothing else?
... owl-db name implies something
<alanr> quantify "large"?
calvanese: profile was created to connect to large databases. we believe it is specifically suited to databases
<alanr> millions, 100s of millions?
<alanr> 10s of billions?
calvanese: also conceptually matches expressivity of databases
zhe: misleading to me because
dl-lite can be provided to other domains
... plus gives users belief dedicated to storing owl
... gives impression only implementable with db, nothing
else
... dl-lite could apply to sparql endpoint as well
calvanese: one point is that its implemented using database technologies
zhe: is this implementation specific?
calvanese: its how the profile
came about
... its tuned to these features
ianh: useful exchange, and what we suspected. owl-db is controversial. any other less controversial names?
<Rinke> Profile names are easily interpreted as denoting disjoint `features'
bmotik: why not 1,2,3 or A,B,C?
<alanr> the only reasonable mnemonic is "R"
ianh: we have reasonable names for EL++ and OWL-R which people are comfortable with. isn't 1,2,3 silly?
bmotik: what's wrong with current names?
ianh: owl dl-lite is too much of a mouthful
alanr: only name with good pneumonic is OWL-R, EL++ is historical and only relevant to small audience
<sandro> +1 get away from history.
alanr: I support getting away from historical names and suggest 1 letter (fairly meaningless) names
<alanr> yes, peter, but for how many others?
<sandro> "DL" is another bad name.
bmotik: owl dl-lite is too much of a mouthful, what about just dl-lite
<alanr> I agree, that DL is another bad name
bmotik: el++ has established itself, it doesn't need the owl prefix
ianh: that may be a step too far
<alanr> OWL-C for OWL-DL (OWL-Complete)
<alanr> OWL-A for (OWL-Anything for OWL-Full)
<Zakim> sandro, you wanted to propose leaving this to marketing
<Rinke> DL-Lite: about assertions, why not OWL-A
sandro: we are worst people to
pick names. someone should subject a marketing department to
this not us
... knowledge of history is an impediment
ianh: another side, the marketing people ask you to explain because they know nothing. so, names they create will depend on who explains them
<Rinke> agree with Sandro, one complaint that came up in my little survey was that people didn't know what the names meant
<ivan> +1 to Rinke
sandro: names should be targeted at people making the purchase decision
calvanese: name is indication, choice will be made on features
<alanr> I was convinced
calvanese: I made several good
arguments for why owl-db is good for dl-lite
... I didn't hear compelling, non-marketing
counterarguments
zhe: why not call owl-r owl-db? oracle is largest database in the world and implements owl-r?
<JeffP> We can call it OWL-Aberdeen
<ivan> JeffP: I would prefer OWL-Amsterdam!
<JeffP> hehe
<Rinke> me too!
ianh: enough of this discussion. owl-db is just too attractive, so probably no one can have it
<sandro> +1 to random city names. :-)
<ivan> rowl, dowl?
<Rinke> howl?
<alanr> who gets OWL-Bagdad?
UNKNOWN_SPEAKER: anyone?
pfps: I don't think anything needs to be done, current status is fine
ianh: current status is that we're using rdf reification
alanr: I'm happy with current reification
<Zhe> second alanr
alanr: as long as triple being reified is included
bmotik: I don't think we should
output triple being reified
... this can be handled in the semantics
<alanr> that's not an argument against. It's an argument that says we can also do it a different way
zhe: conceptually, bmotik is 100%
correct. but with tons of annotations this makes implementers
life difficult
... what's the objection to adding the triple
alanr: yes, what's argument against? this is a divergence from rdf semantics
<alanr> I put a proposal for how to solve this on the email
bmotik: impossible to know when
mapping rdf to ontology if ontology contained axiom or just
annotation of axiom
... I consider sticking with current better solution
+1 to supporting annotation of non-present axioms
<alanr> There is also rdf/xml support for concise reification when it includes the triple
pfps: I don't believe argument that additional processing burden is accurate since it introduces an additional triple to parse
zhe: bmotik, I believe you
proposed solutions via email to some of these problems.
... pfps, oracle believes not including triple will make life
harder
<bmotik> +q to respont to Zhe
alanr: support for concise reification in RDF/XML, but only in some circumstances
<sandro> (er, no, you still need to parse the triples even when not using the RDF/XML trick.)
bmotik: are you proposing we use this special syntax
alanr: if triple is in
serialization, on can put an id on the predicate to indicate
reification
... there is no shorthand for only the reified part
ianh: closing discussion soon
<Zakim> bmotik, you wanted to respont to Zhe
<alanr> no bad ida
<alanr> better to add a special annotation so they are parallel
<alanr> I don't understand
<ivan> me neither
bmotik: one could use following procedure.... if re-ified and non-reified version are present... but this is non-monotonic. question to zhe - if hint that reified triples in RDF/XML should use this shorthand, would that be ok?
<ivan> I do not think we can do that, Boris
<Zhe> sounds good
ianh: take to email, then revisit discussion
ianh: agenda has short list of
things needing attention
... features: 1) rich annotations, 2) nary datatypes
<uli> Bijan isn't here
ianh: no bijan? :( perhaps uli on nary?
uli: what are you after?
ianh: I'd like some comments on schedule?
uli: we could be moving really faster. I won't be around for next two weeks, otherwise I'd say proposal in 1 week
ianh: a concrete proposal for
what should be added to spec?
... but not now?
<alanr> probably depends on what happens next week and the week after too...
uli: depends on this week.
ianh: this is reasonable guesstimate
alanr: we should get quick
check-in on prioritizing things. rich annotations, nary
... how are people on nary? priorities, benefits vs cost of
delaying?
... when do we say it's out?
ianh: is my answer some number of weeks?
<bmotik> -q
<bmotik> +q
alanr: I would like to hear from people. I'd like to hear input.
ianh: is it significant delay worthy?
msmith: I think nary are important and would be prepared to wait some
<uli> "How horrible would you think failing on n-ary be?"
<pfps> i'm prepared to wait forever as long as it isn't more than 15 minutes (thanks Oscar Wilde)
<bmotik> I believe that n-ary datatypes are a high-risk feature
<Achille> we can leave without nary
<alanr> I'm concerned about unknowns with n-aries, and known issues, like difficulty in combinations.
bmotik: adding nary adds a huge burden to developers. some algorithmic issues haven't been resolved and I'm skeptical
msmith: notes Carsten also absent
<uli> good point
<Achille> it is not worth delaying the spec for it
ianh: not time now to get to into the details
<ivan> owl3?
uli: not having any nary support would be regretted later as something we missed
ianh: perhaps we should set some
implementation bar
... 2 implementations to get to rec, correct?
sandro: in general, should only add things for which we think reasonable to there may be two implementations
<pfps> +1 to "at risk"iness
sandro: if we're unsure, that means its at risk
alanr: that doesn't help because there's significant work to get it into the spec
ianh: i agree with that, but the implementation point clarifies just how much expressive power we want to add.
<Zakim> alanr, you wanted to give one more thought
ianh: those wanting it very powerful must weight that against cost of implementing it so that it can proceed
<bmotik> +q
alanr: so far focused on one type of concrete domain extension, < > simple arithmetic
<uli> alanr, can you explain this?
alanr: perhaps allen interval relations instead. I'm taking this up with carsten
<alanr> uli, yes, via email
bmotik: allen interval for time intervals will not solve problems for owl
<alanr> won't solve all time problems for time. But may solve some some time problems
bmotik: nary datatypes won't help
this ...
... (scribe interpret) because they only apply to data
properties on a single individuals (not comparison between
multiple events)
ianh: will ask bijan next week
about this
... also discussion about datatypes in general, what should be
supported. is this going to derail us?
<bmotik> +q
bmotik: thinks we can resolve. we have to resolve
<alanr> I have concerns that this will take time.
bmotik: I don't think solution is difficult
<uli> ;)
<uli> yes
<uli> very
ianh: ISSUE-118 is
languishing
... any champion for this issue?
alanr: I've suggested unnamed and bnodes as alternative constructs
ianh: documents need to be produced. test, ufds
<alanr> ACTION: alan to send email re: suggestions (again) for unnamed individuals *in addition* to bnodes [recorded in http://www.w3.org/2008/06/25-owl-minutes.html#action03]
<trackbot> Created ACTION-164 - Send email re: suggestions (again) for unnamed individuals *in addition* to bnodes [on Alan Ruttenberg - due 2008-07-02].
<alanr> mike, you know about action=raw ? to get raw mediawiki pages?
alanr, no. thanks
msmith: re tests, I'm targeting
f2f3 as a milestone. two parts, the tests, and the
documents
... I'll try to get something to the group before f2f3 on
each
<alanr> mike, see http://svn.neurocommons.org/svn/trunk/product/wiki/get-ncpage-ontology.pl
ianh: none for ufd
pfps: I think bijan is working on primer
<JeffP> bye
<ivan> bye
ianh: no additional business, adjourn
<Zhe> thanks
<uli> bye bye
<Ratnesh> bye
<sandro> msmith, I put some notes about scribing here: http://www.w3.org/2007/OWL/wiki/Scribe_Conventions#After_scribing_.28New_Style_Minutes.29
<IanH> I'm hoping that Turkey wins on penalties :-)
<alanr> e.g http://sw.neurocommons.org/cgi-bin/get-ncpage-ontology.pl?page=CommonsPurl:Record/Ncbi_gene§ion=purlRdf
<uli> I hope that they have great game
This is scribe.perl Revision: 1.133 of Date: 2008/01/18 18:48:51 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/cerned/concerned/ Found ScribeNick: msmith Inferring Scribes: msmith WARNING: Replacing list of attendees. Old list: Peter_Patel-Schneider +1.202.408.aabb +0186527aacc Ivan Sandro Ratnesh Zhe msmith +1.518.276.aadd baojie bmotik IanH uli +0122427aaee Achille JeffP bcuencagrau Alan calvanese Rinke New list: Peter_Patel-Schneider +1.202.408.aabb +0186527aacc Ivan Sandro Ratnesh Zhe msmith +1.518.276.aadd baojie bmotik IanH uli Default Present: Peter_Patel-Schneider, +1.202.408.aabb, +0186527aacc, Ivan, Sandro, Ratnesh, Zhe, msmith, +1.518.276.aadd, baojie, bmotik, IanH, uli Present: Peter_Patel-Schneider +1.202.408.aabb +0186527aacc Ivan Sandro Ratnesh Zhe msmith +1.518.276.aadd baojie bmotik IanH uli WARNING: No meeting title found! You should specify the meeting title like this: <dbooth> Meeting: Weekly Baking Club Meeting Agenda: http://www.w3.org/2007/OWL/wiki/Teleconference.2008.06.25/Agenda WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Got date from IRC log name: 25 Jun 2008 Guessing minutes URL: http://www.w3.org/2008/06/25-owl-minutes.html People with action items: alan bmotik2 calvanese[End of scribe.perl diagnostic output]