RIF Telecon 12-Feb-08 -- 04 Mar 2008

<ChrisW> Meeting: RIF Telecon 4-Mar-08

<ChrisW> leora, can you scribe today?

yes

<ChrisW> Scribe: LeoraMorgenstern

Admin

Sandro: There was sort of a telecon last week: Christian, Harold, and I summarized the F2F; it was not something that would count as an official telecon.

csma: no scribe, no minutes, no meeting.

<ChrisW> Feb 12 Telecon Minutes: http://lists.w3.org/Archives/Public/public-rif-wg/2008Feb/att-0094/12-rif-minutes.html

<ChrisW> PROPOSED: Accept 12 Feb Telecon minutes

<ChrisW> RESOLVED: Accept 12 Feb Telecon minutes

<ChrisW> http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/att-0005/2008-02-19-rif-minutes.html

PROPOSED: Accept 19 Feb Telecon minutes

RESOLUTION: Accept 19 Feb Telecon minutes

Chris: Any agenda amendments?

No agenda amendments

Chris: At next week telecon, Europe and U.S. will be out of sync.
... US will be on summertime, but Europe switches 3 weeks later.

csma: Do we change the reservation in Europe or in US?

sandro: might cause problems to change in US.
... reservation is with respect to US time.

Chris: We always keep to the US times. So next week's telecon will be at 4 PM in Central European Time, and 3 PM in British Time.

F2F10

Chris: Minutes for F2F9 are up and linked to on wiki page. We will vote on them next week.
... On to F2F10.
... There was a web poll, and it seemed that the preference was clearly for Galway, Ireland, May 26-28.
... 3 day meeting, because it will be the last RIF F2F meeting. All the things that need to get done have to get done at that meeting.

<IgorMozetic> http://www.w3.org/2002/09/wbs/38457/f2f10dates/results

<csma> http://www.w3.org/2002/09/wbs/38457/f2f10dates/

<josb> March? => May

<ChrisW> May, yes

Chris: Okay, decided that next F2F will be May26-28 in Galway.

<sandro> RESOLVED: F2F10 Galway, Ireland, 26-28 May (Mon-Wed)

<ChrisW> ack ??

<sandro> ack ??P26

F2F9 and action review

Chris: Any other discussion on F2F10?

None.

Chris: Working Group's main page contains table, put up by Sandro, of documents to be done, along with the schedule.
... We closed issue 44, a grain of sand that had been irritating us, regarding uniterms. That was done on the first day.
... Did clean up on the specs of documents (get links, numbers).
... Another major resolution: Decided how to handle errors in BLD.
... Went with third approach to handling errors: let implementors decide whether to return true or false for presence of errors.
... spec is requiring a guard predicate.
... Another resolution: We discovered BLD had reification. (Indeed, it was intentional.) We decided to take it out of BLD; framework, however, still permits you to write a dialect that has reification.
... Now, on to action review:

There was Action 442 on Harold to make a proposal for adding metadata to the BLD syntax.

Harold: will get it done 2 weeks from now.

csma: but that will be after freezing date of document.
... since freezing date is 11 of March.

Due date for Harold has been changed for 18 of March.

Also Action 441 on Harold to add IRIs to presentation syntax.

Jos: In proposal sent to the mailing list, metadata is included as well as IRIs.

But Action 442 is not obsolete, because it still hasn't been added to document.

So 441 and 442 are continued. May become obsolete.

Adrian not here. Jeff Pan (test cases for Rif) not here.

Action 438 on Jos is pending discussion.

chris: Michael (action 437) add built-ins to semantics of BLD and FLD

michael: this is ongoing
... will have it done within a week, maybe.
... will be going into next version

Chris: did Axel finish list of built ins including typechecking and casting

<josb> I'd say the lists are not finished.

Chris: So action 436 is pending discussion.

Sandro: action 435 is continued.

<josb> (E.g., casting functions are not included at all )

csma: action 434 is ongoing, due on Friday.

michael: action 433, to move section 2.0.9 to appendix: Possible to have one section about deriving BLD from FLD, and then one section for the semantics, and that will be clearer ---
... once this is done, it can be determined whether it is better to move that section into an appendix.
... that will also be done March 11.

chris: action 431 can be deleted.

Action 432 is continued.

michael: regarding actions 430 and 433: these are connected: sections will be combined.
... did not have a chance to look carefully at action 429, which depends on action 428, by Axel (and Harold). 428 has now been done, so will be doing 429 now.
... I'll do it all together this week.
... same for actions 426 and 427.

<Harold> Michael, Jos and all, re lists, there was only an open choice about two possible semantics: I'm fine with both. So, if no one wants the original 'pair' semantics, or has a problem with the 'nested-interpretation semantics', then let's go for the latter.

correction: action 426 is obsolete and now closed.

<josb> Okay, let's do that

discussion on action 425, to make sure that BLD requires explicit quantification. What exactly was this action?

josb: in BLD document spec, there were two types of rules, one with and one without quantifiers, and it said explicitly that rules without quantifiers were allowed, so that has to be removed.

michael: wasn't that just a bnf issue?

josb: no, not just a bnf problem.
... there should only be one type of rule, one with quantifiers.
... one without quantifiers should be discarded, because we decided that all quantifiers must be explicit in BLD.

michael: again, I'll be looking at all of these issues during this coming week.

Harold: Action 423 depends on actions by Axel and Michael.

(get Harold's remarks: can't hear him.)

Harold: Do we require a guard for every built-in?

Chris: no, not required, just recommended.

Harold: can't do it earlier than 10 of March.

csma: okay, that just all right, time wise.

Adrian had action 413, but is not here.

csma: Action 413 was done and discussed during f2f, so is now done.

chris: action 406 is done.
... action 405 on jos was done and discussed, so can now close it.

harold: action 404, to update BLD syntax/semantics to reflect resolution on lists, was discussed a bit on the IRC.
... There's a choice to make between two semantics for lists, and we have to decide on which one.

chris: that needs to be on the agenda for next week.

Actions 384 and 400 are pending discussion.

<josb> Harold, <josb> Okay, let's do that

<Harold> OK.

<DaveReynolds> I agree, collation issue is certainly not critical.

Sandro: action 382 can be dropped.

So action 382 is closed.

Action 378 is still pending discussion.

sandro: Action 373 is done.

chris: Actions 152, 253, 274, 305, 359, and 361 are all continued.

Liaison

<josb> no

chris: Jos, anything new regarding OWL?

(see Jos's answer above.)

DTB

Chris: new document (on data types and built ins) is up on the wiki. Harold, what's the status?

Harold: <can't hear anything>

<Harold> E.g.: 2.1.1.3 & proposal (Axel's proposal)

josb: I saw a version, made comments, but haven't seen comments addressed.

chris: Harold said Axel addressed the comments.

josb: not so.

<Harold> http://www.w3.org/2005/rules/wiki/DTB#.26_proposal_.28Axel.27s_proposal.29

josb: for example, I had comments about how built ins are defined, etc.

chris: why were there new things, there?

josb: the existing document was insufficient --- wrt definitions, etc.

csma: I understood that wg decided to use built-ins for uniterms.

<josb> my comments: http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/0010.html

csma: but here I see all the other proposals are still listed.
... in other words, I thought the question was settled, but clearly that's not the case.

chris: so document hadn't been properly updated.
... so it looks like proposal 2.1 can go.

(note to self: check the numbering)

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2008Feb/att-0094/12-rif-minutes.html#item08

chris: so section 2.1.1 can go, and 2.1.1.3 --- all subsections on other proposals.
... can go

josb: indeed, my comments were that these sections were part of language specification

(note to self: check josb's email.)

<Hassan> march 11

chris: we want to freeze this by next week.
... but harold, you're travelling and Axel is not here. Is march 11 a realistic date to freeze this document?

harold: no.

csma: Let's freeze March 18, and give only one week for review.

chris: we did not actually decided on a schedule for next working draft.
... let's freeze by march 14 and have 11 days for review.

csma: people should send comments as soon as possible, so frozen version of doc will already include most of comments.

FLD

chris: plan is to freeze document next week. Michael is optimistic that this can happen.
... Michael, anything has come up since f2f that we should discuss?

michael: only thing is that I need time to work on this. The two documents are closely related, and I have to work on them together.

<ChrisW> http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/0001.html

<josb> but it's so profitable!

Josb and Michael: email discussion

michael: in FLD, have to talk about formulas; in BLD, want to talk about rules.
... way to restrict things is that BLD is restriction of FLD grammar.

josb: Restriction goes the other way round.

<Hassan> A rule is not a formula for Business Rules systems

It appears that not all points brought up in email discussion by Michael and Jos have been responded to by other party.

michael: discrepancy between rule and formula, and spec of production rules not looking right.
... rule content looked redundant.

josb: Yes, I pointed that out too.
... not sure how far BLD grammar was restriction of FLD grammar.
... from my point of view, not so important, as long as clear that syntax of BLD is restriction of syntax of FLD.

<Harold> Current Version (http://www.w3.org/2005/rules/wiki/BLD#EBNF_for_RIF-BLD_Condition_Language)

<Harold> CONDITION ::= 'And' '(' CONDITION* ')' |

<Harold> 'Or' '(' CONDITION* ')' |

<Harold> 'Exists' Var+ '(' CONDITION ')' |

<Harold> COMPOUND

<Harold> COMPOUND ::= Uniterm | Equal | Member | Subclass | Frame

<Harold> Uniterm ::= Const '(' (TERM* | (Const '->' TERM)*) ')'

<Harold> Equal ::= TERM '=' TERM

<Harold> Member ::= TERM '#' TERM

<Harold> Subclass ::= TERM '##' TERM

<Harold> Frame ::= TERM '[' (TERM '->' TERM)* ']'

<Harold> TERM ::= Const | Var | COMPOUND

<Harold> Const ::= LITERAL '^^' SYMSPACE

<Harold> Var ::= '?' VARNAME

<Harold> Prohibiting Reification (F2F9 Day 1 Resolution)

<Harold> CONDITION ::= 'And' '(' CONDITION* ')' |

<Harold> 'Or' '(' CONDITION* ')' |

<Harold> 'Exists' Var+ '(' CONDITION ')' |

<Harold> ATOMIC

<Harold> ATOMIC ::= Uniterm | Equal | Member | Subclass | Frame

<Harold> Uniterm ::= Const '(' (TERM* | (Const '->' TERM)*) ')'

<Harold> Equal ::= TERM '=' TERM

<Harold> Member ::= TERM '#' TERM

<Harold> Subclass ::= TERM '##' TERM

<Harold> Frame ::= TERM '[' (TERM '->' TERM)* ']'

<Harold> TERM ::= Const | Var | Uniterm

<Harold> Const ::= LITERAL '^^' SYMSPACE

<Harold> Var ::= '?' VARNAME

<Harold> Reintroducing Atom/Expr Distinction (F2F9 Day 2 Discussion)

<Harold> CONDITION ::= 'And' '(' CONDITION* ')' |

<Harold> 'Or' '(' CONDITION* ')' |

<Harold> 'Exists' Var+ '(' CONDITION ')' |

<Harold> ATOMIC

<Harold> ATOMIC ::= Atom | Equal | Member | Subclass | Frame

<Harold> Atom ::= Const '(' (TERM* | (Const '->' TERM)*) ')'

<Harold> Equal ::= TERM '=' TERM

<Harold> Member ::= TERM '#' TERM

<Harold> Subclass ::= TERM '##' TERM

<Harold> Frame ::= TERM '[' (TERM '->' TERM)* ']'

<Harold> TERM ::= Const | Var | Expr

<Harold> Expr ::= Const '(' (TERM* | (Const '->' TERM)*) ')'

<Harold> Const ::= LITERAL '^^' SYMSPACE

<Harold> Var ::= '?' VARNAME

chrisw: several things discussed in f2f are reflected in Josb's new bnf: has metadata, has iri's, etc.

chris: also had some discussion based on sandro's attempt to do tooling based on xml syntax.
... previously, syntax didn't distinguish between predicates and functions.
... although earlier versions did this.
... and indeed, we are considering bringing this distinction back.

josb: the BLD and FLD grammars now do make this distinction, and it turned out to be not problematic, so there is no reason not to make the distinction.

csma: Why do you have to change rules and rule content in order to separate predicates and functions?

chris: that was to get metadata.

csma: if we want to map bnf on the xml syntax, then the way you do it is problematic, because we do not have separate productions for built-in and non-built-in forms of predicates and functions.
... that is, there is only one production for built in functions and predicates
... and only one production for non-built-in functions and predicates.
... we would need a production for each xml form.

chris: they can't share the uniterm?

<Harold> The production Predicate ::= UNITERM | 'Builtin ( ' UNITERM ' ) ' does not mark a builtin call (say with Exterm).

csma: but they don't have the xterm.

<josb> http://www.w3.org/2005/rules/wiki/BLD#Translation_Between_the_RIF-BLD_Presentation_and_XML_Syntaxes

josb: link above shows how you do it.

harold: we need to separate different issues.
... one issue is getting rid of reification.
... second issue is getting rid of universal terms (??)

(note to self: email Harold and get comments, none of which I heard.)

harold: in terms of getting rid of uniterms: difference between atoms and expressions.
... once we have made this difference, cannot introduce uniterms anymore.
... that was the resolution or decision, a year ago.

michael: we need a bnf from which it can all be derived.

josb: Asks harold to respond to his grammar, and specify what's wrong with it.

chris: harold doesn't think you need to split off uniterms from existing syntax.

<Harold> We had a decision, approx. a year ago, to start with unified tags, and in later dialects refine them, rather than start with differentiated tags, and later try to re-unify them.

chris: again, need to be able to distinguish in syntax itself, difference between functions and predicates, and the second issue that needs to be fixed in the syntax, is to make sure that there's no reification.
... and we need metadata
... and we need to be able to refer to IRIs

<csma> and we need it by next week!

chris: Jos's proposal addresses all of these.
... but the existing syntax doesn't.
... we will discuss more next week, and hopefully come to a conclusion.

<Harold> We applied this to the unified Uniterm tag rather than keeping our earlier Atom and Expr tags.

BLD

chris: Harold, can you send your comments as an email, since some didn't understand them?

In fact, some didn't even hear them.

chris: Well, we've already been discussion BlD. Anything else to discuss on BLD?

Dave: Did the various xml syntax issues-- things like IRIs, etc --- get sorted out?

chris: that is one of the things that Jos's sytnax proposal addressed.

<csma> unmute me

Dave: no, it's a different issue: in the proposed syntax, various xml issues were not addressed.
... I mentioned them in my comments on the proposal.
... Did those get sorted out?

<Harold> I agree with Dave, this are different issues. And issues should not be compounded.

csma: We didn't really discuss actual xml syntax
... in itself,
... These issues will become critical path shortly

Dave: yes, necessary for implementation.

Chris: We should go over that next week as well.

<csma> ACTION: to CWelty to add XML syntax on the agenda for next week [recorded in http://www.w3.org/2008/03/04-rif-minutes.html#action01]

<trackbot-ng> Sorry, couldn't find user - to

Harold: Something about modules. (can't hear most of this.)

<csma> ACTION: to christopherwelty to add xml syntax on the agenda for next week [recorded in http://www.w3.org/2008/03/04-rif-minutes.html#action02]

<trackbot-ng> Sorry, couldn't find user - to

harold: not talking aobut rif-level modules, but about xsd-level modules.

csma: (2) importing rules sets
... (3) moduloes

(csma was summarizing harold's points. Get first point.)

<Harold> http://www.w3.org/2005/rules/wiki/BLD#Appendix:_Specification

Harold: Jos's language is now one language that covers conditions and rules.

<csma> ACTION: to chriswelty to add action-404 on agenda next week [recorded in http://www.w3.org/2008/03/04-rif-minutes.html#action03]

<trackbot-ng> Sorry, couldn't find user - to

SWC

Harold: I think we should separate concerns, not merge the issues.

Chris: regarding owl and rdf compatibility:
... were hoping to freeze next week.

josb: It is on schedule if BLD spec is on schedule.
... 3 days after BLD spec is frozen, this can be frozen.

chris: michael will finish and tell sandro, then sandro will freeze and notify.
... jos, what changes were you depending on?

michael: changes involve moving sections around, etc. and hoping that everything works perfectly. One shouldn't be too optimistic.

Dave: WRT semantic compatibility document, was there any discussion about the difference between the OWL and RDF approaches to compatibility.

Chris: Yes, lots of discussion, and realistically, there will be difficulties between OWL and RIF given that there are 3 different versions of OWL.
... Are you mostly concerned with using classes as predicates vs. using frames?

Dave: yes, this makes the problem worse. You have to decide wihether it's OWL-DL or OWL-full before determining how to translate.
... Do you translate it as predicates or translate it as frames?

<sandro> I want the hit the next person who says "RIF is just an interchange format"

Chris: But RIF is just an interchange format so presumably the one doing the translating knows which version they support.

Josb: Perhaps Dave has a misconception of rif syntax for owl rules
... You still use the same RIF rules, just give them a different semantics.
... Granted you must decide which semantics to use, but the same is true when you are just using OWL.

chris: does that address your major issue, Dave?

Dave: I don't understand how an implementor can implement that at this point.
... I don't understand how the semantic equivalence is supposed to work.

<sandro> Jos: When you write rules against OWL, you pick whether you want to use DL-style semantics or Full-style semantics.

Chris: Dave, can you be more specific about the problem?

<ChrisW> leora, can you stay?

Dave: If it was clear how to translate unary and binary predicates, it would be easier.

Josb: can only do that for OWL-DL, not Owl-Full.
... Owl-Full: triple semantics.
... Owl-DL predicate semantics.
... distinction between classes and properties.
... triples equivalent to unary and binary predicates.

Job: already pointed out in specification: more entailments in OWL-full semantics than in OWL-DL semantics.

Josb: and this is true in RIF as well.
... Suggestion --In RIF OWL-DL combinations, disallow users to write unary and binary predicates
... don't allow users to use two different syntaxes for the same thing.

<DougL> I have to leave now but am interested in the outcome of this discussion. Thanks.

Josb: Only disallow it in OWL-DL.
... But allow it for OWL-full and RDF.

Dave: I think I'm okay with that.

Josb: Will do my best to explain that better in the document.

- DRAFT -

RIF Telecon 12-Feb-08

04 Mar 2008

Attendees

Contents

Admin

F2F10

F2F9 and action review

Liaison

DTB

FLD

BLD

SWC

Summary of Action Items

Scribe.perl diagnostic output