W3C

- DRAFT -

RIF Telecon 11 Dec 2007

11 Dec 2007

Agenda

See also: IRC log

Attendees

Present
Mike_Dean, Harold, ChrisW, josb, PaulaP, Stella_Mitchell, IgorMozetic, Hassan_Ait-Kaci, csma, LeoraMorgenstern, markproctor, Gary_Hallmark, Michael_Kifer
Regrets
Axel
Chair
Chris Welty
Scribe
Mike Dean

Contents


 

 

<ChrisW> Scribe: Mike Dean

Admin

<markproctor> is there a belgium number I can call?

<markproctor> I'm at a conference, so no landline

<markproctor> and international calls to the US on a mobile are very expensive.

<csma> Mark, there is a number in France

<markproctor> I can get access to a belgium mobile, if there is a belgium number.

<markproctor> no belgium one?

<markproctor> if not France will have to do

<csma> +33 4 89 06 34 99

<markproctor> where do I find the details? this only lists the US number - http://www.w3.org/2005/rules/wg/wiki/Telecons

<ChrisW> we should update that

<markproctor> ok, thanks.

<markproctor> I'll just go and find that belgium mobile and call in- hopefully calling france from belgium on a mobile isn't too much.

<ChrisW> i updated the wiki page with the French number, Christian you should update the email agenda template

<scribe> Scribe: Mike Dean

<scribe> ScribeNick: mdean

<csma> ACTION: christian to update the agenda template with Zakim french phone number [recorded in http://www.w3.org/2007/12/11-rif-minutes.html#action01]

<rifbot> Created ACTION-394 - Update the agenda template with Zakim french phone number [on Christian de Sainte Marie - due 2007-12-18].

ChrisW: postpone approval of last week's minutes since they've only been out for a day

<markproctor> +MarkProctor

<scribe> Chair: Chris Welty

<markproctor> damn forgotten how to register a new number

action 393 on axel is complete

action 392 on sandro is complete

action 390 on Christian is complete or irrelevant

action 389 on jos is complete

action 386 on Chris is complete: jos and Mike Dean are RIF members of OWL Task Force - Peter and Uli Sattler are from OWL WG

scribe: may want to pick 1 more from each side

jos: 2 may be preferable

ChrisW: ensure all stakeholders are accounted for

josb: interested in OWL Full as well as OWL DL

action 387 on leora is continued

action 384 pending discussion

Christian updating actions

<markproctor> I'm in two conference calls at the same time

<markproctor> so do say my name twice, incase I can't follow that part of the call - to get my attention.

Liaison

<ChrisW> http://www.w3.org/2005/rules/wg/wiki/Response_to_PPS1

ChrisW: draft of email response

<josb> don"t understand

josb: partial response to question 14
... semantics is same as RDF - syntax is different - michael might justify why

ChrisW: may make second pass
... still need to respond to Peter's second message

<Harold> Re 'Liaison', does someone know anything about the results of the OWL Manchester f2f?

<scribe> ACTION: Harold to review response to Peter Patel-Schneider [recorded in http://www.w3.org/2007/12/11-rif-minutes.html#action02]

<rifbot> Created ACTION-395 - Review response to Peter Patel-Schneider [on Harold Boley - due 2007-12-18].

F2F9

<IgorMozetic> http://www.w3.org/2002/09/wbs/38457/f2f9options/results

ChrisW: not a lot of responses to survey earlier

Christian: 18 answers currently

ChrisW: let's go with Paris

Christian: will provide dinner

Feb 21-22

dates decided last week

RESOLUTION: F2F9 at ILOG in Paris Feb 21-22

<ChrisW> Mike, that's not actually a resolution but a chair decision

Christian: need to book hotels through ILOG - will send email to list

BLD - Issue 40

<csma> Michael, are you joining?

<ChrisW> http://www.w3.org/2005/rules/wg/wiki/List_of_BLD_built-ins

ChrisW: anyone who doesn't like option C modulo syntax changes in email?

<PaulaP> +1 for Option C

Christian: logic functions would be different in what way?

<josb> right

ChrisW: external calls have their own syntax

<Harold> Re option c, The Presentation Syntax does not use commas but whitespace as separators:

<Harold> &fn:dateTime( "2006-08-15"^^xs:date, "12:30:45-05:00"xs:time )

<Harold> -->

<Harold> &fn:dateTime( "2006-08-15"^^xs:date "12:30:45-05:00"xs:time )

<josb> http://lists.w3.org/Archives/Public/public-rif-wg/2007Dec/0039.html

at end of message

<Harold> Also, I think we should not 'use up' the "&" special character for this purpose; "fn:" is clear enough.

<csma> I would prefer <Apply>...</Apply> instead of <Exterm>

<ChrisW> PROPOSED: Go with Axel's option C, that is using a special syntax to distinguish evaluated predicates from logical functions

<josb> And (plus(z,1,1) z>y)

<Harold> Michael, do you mean a new predicate greaterThanSum(y 1 1)?

<csma> harold, this too complex

Michael: allow only total functions, not partial

<Harold> Christian, I didn't suggest this :-)

Christian: otherwise define binding patterns

<Harold> However, in F2F8 we tended strongly towards having only builtin predicates.

<Harold> This was because we took out Equal from the (future) Core.

<Harold> Without Equal, we can no longer define user-defined functions,

<Harold> so it makes little sense to have builtin functions.

ChrisW: resolution is about syntax

<ChrisW> RESOLVED: Go with Axel's option C, that is using a special syntax to distinguish evaluated predicates from logical functions

<Hassan> Makes sense now ... Thanks Igor!

RESOLUTION: Go with Axel's option C, that is using a special syntax to distinguish evaluated functions/predicates from logical functions/predicates

Harold: removing Equal may preclude user-defined functions

Christian: wants to allow user-defined functions

<GaryHallmark> equality is in BLD, just not in Core

ChrisW: syntax discussion

<ChrisW> ATOMIC ::= Uniterm | Equal | ExtTerm

<ChrisW> TERM ::= Const | Var | Uniterm | ExtTerm

<ChrisW> ExtTerm ::= '&' Const ' ( ' TERM* ' ) '

Christian: probably not ready to resolve this now

<josb> ExtTerm ::= ' BuiltIn( ' Uniterm ' ) '

<GaryHallmark> I don't agree. Most PR will be all builtins

<IgorMozetic> but this might make confusion with higher-order predicates

ChrisW: add to Wiki page then discuss next week

Gary: prefer simpler presentation syntax
... prefers & or something similarly concise

<markproctor> for "builtins" we talking about the first order logic conditional elements? exists, not, forall, collect ?

<markproctor> or are you talking about possible functions used inside test/eval nodes?

<markproctor> typically what goes in a test/eval node is black box.

<csma> no, we are talking about procedural attachments

<markproctor> ok so test/evals

<csma> right

<Michael_Kifer> I agree with Gary

<PaulaP> I can do that

<markproctor> so many engines come with a selection of functions, but you can easily add more.

<josb> how about ExtTerm ::= ' &[ ' Uniterm ' ] '

<PaulaP> ok

<csma> same possible confusion with embedding, Jos

<PaulaP> ok

<PaulaP> http://www.w3.org/2005/rules/wg/wiki/List_of_BLD_built-ins

<ChrisW> http://www.w3.org/2005/rules/wg/wiki/List_of_BLD_built-ins

<Harold> Re builtin functions as well as builtin predicates, we should not introduce the **redundancy** of having all/many of the (arithmetic, ...) builtins TWICE, functionally as well as logically.

<scribe> ACTION: PaulaP to update BLD list of builtins for new external calls [recorded in http://www.w3.org/2007/12/11-rif-minutes.html#action05]

<rifbot> Sorry, couldn't find user - PaulaP

BLD - Issues 41 and 43

Michael: charter addresses frame syntax
... to accommodate round-tripping of frame-based languages
... Core as profile of BLD
... defining dialect is a lot of work, but profile (restrictions) are easy
... if we don't include subclass and membership in BLD, then no reason to include frames

Christian: data model should be passed out-of-band
... this is the argument against

<josb> Sorry, I have to go now

<josb> I am not against getting rid of frames in core

<josb> bye

<GaryHallmark> argument for # and ##: you can exchange more rules without understanding details of data models

Christian: suggest BLD++ with membership and subclass
... will need mechanism to extend dialects

Michael: such a mechanism will be very complicated

<GaryHallmark> argument against # and ##: you have to "bend" your data model to conform to RIF's notion of # and ##

Michael: BLD should be technically complete

Christian: will also have to discuss equality

<GaryHallmark> hand in hand

ChrisW (not as chair): would like membership but not classification

<markproctor> was there a URL to provide the definition difference between the two?

<Hassan> I have no clue ???

straw poll indicates that the 2 should be treated together, although jos and Dave Reynolds aren't on call

<markproctor> so no wiki page explaining the difference?

<Hassan> To me they are the same thing ... If you can do one, you can do the other. Membership = subtype + singleton-denoting-set

<GaryHallmark> +1 with Michael, too much design by committee is not good

ChrisW: doesn't make sense to discuss this now

<csma> they are defined in BLD

BLD - Issue 45

<ChrisW> http://www.w3.org/2005/rules/wg/track/issues/45

<markproctor> someone want to do an action item to define classification/membership in the wiki?

<ChrisW> agendum Issue 45 Lists

<markproctor> no isn't it issue 44?

<csma> a # b is: instance a is a member of class b

<markproctor> Issue 44 (named arguments Uniterm) [13]

<csma> a## b is cals a is a subclass of class b

ChrisW: don't yet have a proposal for lists

Christian: lists in logic vs. list as type

<GaryHallmark> if you have logic functions, you have logic lists

<Harold> http://www.w3.org/2005/rules/wg/wiki/Core/List_Constructor

Harold: OWL 1.1 sequence construct
... mapped to RDF

ChrisW: just syntactic sugar for rdf:List

<GaryHallmark> need list builtins as well for dialects not supporting logical functions

ChrisW: does OWL 1.1 allow nested sequences?

Harold: don't see any reason why it shouldn't

Gary: won't work so well for production rules
... need builtins to access lists

<Harold> Gary, lists are part of Horn logic, not Datalog.

<markproctor> peeps we didn't discuss 44?

<csma> Michael, I think the easy way out for Core, BLD, BLD++ (or Core, BLD--, BLD) is to have it both way: (1) we keep the BLD document; (2) we specify CORE, BLD-+, BLD+- as restrictions against the complete spec; (3) we call each of them a dialect and) we change the title of the document accordingly. That works because they are specified in the same document, so there is really no difference between a dialect and a profile.

<Harold> So, we could have them only in BLD, not in the Core.

AOB

ChrisW: end early

<markproctor> Issue 44 (named arguments Uniterm) [13]

<markproctor> we didn't discuss that right?

<Harold> But why not have builtins for opaque lists already in the Core.

Hassan will scribe next week

<Harold> ('opaque' in the sense of arrays)

<csma> Mark, no, it will be included in the profile/dialect discussion

ChrisW: adjourned

<PaulaP> bye

<markproctor> ah ok

<markproctor> ok peeps

<markproctor> see ya later

Summary of Action Items

[NEW] ACTION: christian to update the agenda template with Zakim french phone number [recorded in http://www.w3.org/2007/12/11-rif-minutes.html#action01]
[NEW] ACTION: Harold to review response to Peter Patel-Schneider [recorded in http://www.w3.org/2007/12/11-rif-minutes.html#action02]
[NEW] ACTION: PaulaP to update BLD list of builtins for new external calls [recorded in http://www.w3.org/2007/12/11-rif-minutes.html#action05]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.128 (CVS log)
$Date: 2007/12/11 17:21:01 $