RIF Telecon 11 March 2008 -- 11 March 2008

<Harold> Hi Doug, Should we refer to CycL?

<DougL> Hi, sure.

<Harold> Doug how? (I found something online, but maybe you have more precise ref)

<DougL> The wikipedia page for CycL references the CycL syntax document (near the bottom)

<Harold> OK.

Admin

ChrisW: Any objections to accepting the minutes from F2F9?

<ChrisW> http://www.w3.org/2005/rules/wiki/F2F9_Minutes

<ChrisW> RESOLVED: accept F2F9 Minutes

ChrisW: No minutes from March 4th yet

Leora: I just now sent out the minutes from March 4th

ChrisW: Any agenda ammendments?

Christian: Jos also wanted to discuss appendix of SWC (Semantic Web Compatibility) document

ChrisW: OK, we'll talk about that during the publication plan

F2F10

ChrisW: Any news on F2F10? Axel (host) is not here.

ChrisW: F2F10 will be at DERI Galway on May 26-28 (a 3 day meeting)

<csma> ACTION: Axel to update the F2F10 wiki page [recorded in http://www.w3.org/2008/03/11-rif-minutes.html#action01]

<trackbot-ng> Created ACTION-443 - Update the F2F10 wiki page [on Axel Polleres - due 2008-03-18].

Action review

ChrisW: Action-423 is pending discussion

<Harold> ACTION-423: http://www.w3.org/2005/rules/wiki/index.php?title=BLD&diff=526&oldid=513

Harold: The rest of my actions are continued

Sandro: Action-435 (request namespace for functions and operators)
... it's turning out to be harder than expected. I need help from the working group
... I have been in touch with xquery+xpath WGs

Christian: change due date for Action-434 to March 21st

Liason

ChrisW: Christian, any news from the OMG meeting?

Christian: The only thing that might be of interest to this group is that there is a request for proposals on an SVBR vocabulary for date and time that is aligned with OWL and UML

ChrisW: Jos, MikeD, any news from OWL task force?

<josb> no

MikeD: No news

ChrisW: I understand that there is work going on in the OWL WG to consider a blessed (recommended) fragment of OWL for DLP, it is an Oracle initiative.

<Harold> DLP is the intersection of Horn logic and Description Logic.

<sandro> Zhe (Alan) Wu, at Oracle

ChrisW: GaryHallmark, do you know about this?

GaryHallmark: No

MikeD: I will attend the OWLED workshop in early April

ChrisW: Please bring the SWC (Semantic Web Compatibility) document to their attention and solicit feedback

Issue 40 (builtins)

ChrisW: at F2F10 we pretty much agreed on builtins, but in the documented issue there is one item left open, about order of the arguments. Last week, I sent an email with a proposed resolution for this item.

<ChrisW> PROPOSED: BLD builtins are not sensitive to order as they are in query languages and production rules (closing issue-40).

Christian: I have no objection to that resolution, but I wonder what it means that they are sensitive to order "as they are in"...

<ChrisW> PROPOSED: BLD builtins are not sensitive to order

Harold: If you call a builtin before all arguments are bound, you can have a problem in some implementations

Christian: In RIF all bindings are done outside of the rule, so we would not have this problem

<Harold> PROPOSED: BLD builtin calls are not sensitive to order of conjunctions

Harold: Is the above wording OK with you, Christian?

Christian: Yes, even the original wording was fine, but just might be a little confusing

<ChrisW> PROPOSED: BLD builtins are not sensitive to order of evaluation

ChrisW: Any objections to the above proposal? ... none

<ChrisW> PROPOSED: BLD builtins are not sensitive to order of evaluation

<MichaelKifer> +1

<DougL> +1

<josb> +1

<Harold> +1

<Hassan> 0

<IgorMozetic> +1

<sandro> +1

<LeoraMorgenstern> +1

<ChrisW> RESOLVED: BLD builtins are not sensitive to order of evaluation (closing issue 40)

<csma> do you have some wine to celebrate?

<csma> ACTION: cwelty to close issue 40 [recorded in http://www.w3.org/2008/03/11-rif-minutes.html#action03]

<trackbot-ng> Created ACTION-444 - Close issue 40 [on Christopher Welty - due 2008-03-18].

Action 404 (update lists syntax and semantics)

<ChrisW> http://www.w3.org/2005/rules/wg/wiki/Core/List_Constructor-alt

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

ChrisW: We agreed on syntax, but not on semantics yet
...Above are 2 proposals that have the same syntax, but different specification of semantics

<csma> PROPOSED: Approve Michael's alternative proposal on lists [6] and

<csma> update FLD+BLD syntax/semantics accordingly to reflect that and the

<csma> previous resolution on lists

Harold: I have no real preference between the two. I think we should use MichaelK's alternative proposal since no one objected to it
... I think on one level the semantic interpretation is more complicated in MichaelK's proposal
... it is kind of unusual, but it seems to work

ChrisW: Can you clarify?

<Harold> These functions are required to satisfy the following: Itail(a1, ..., ak, Iseq(ak+1, ..., ak+m)) = Iseq(a1, ..., ak, ak+1, ..., ak+m).

Harold: This leads us into the realm of semantic description that is more expressive than in my original proposal

<josb> yes

ChrisW: Any other discussion on this?
...Are people ready to accept MichaelK's ("alternative") proposal for list semantics?
...Does anyone feel uncomfortable accepting the semantics of the "alternative" proposal?

<LeoraMorgenstern> So, we are voting for one of the two pages?

<ChrisW> PROPOSED: Approve Michael's alternative proposal on lists and update FLD+BLD syntax/semantics accordingly to reflect that and the previous resolution on lists

ChrisW: Does anyone object to the above proposal?

<LeoraMorgenstern> I'm confused. Which wiki page are we voting for?

<Hassan> Why not use the standard free algebra style of semantics?

Hassan: I think it is overly complicated. There are standard semantics for lists everwhere, why are we reinventing the wheel?

Harold: To keep it n-ary

Hassan: That is just syntax

Harold: First step was to eliminate pairs from the syntax, and then we eliminated pairs from the semantics too
... Hassan, how would you deal with rest variables?

Hassan: Just a logic variable

<Harold> Itail deals with rest variables.

MichaelK: We have a model theory so when we introduce a new kind of term we have to define the interpretation of this new kind of term in the model theory
... Hassan, you have to be specific about what you are proposing

Hassan: Use standard semantics and syntactic sugar transformation. I don't object to the current proposal, I am just saying my opinion

<Harold> Direct treatment of 'Seq(' TERM+ ` | ` TERM ')'.

<Harold> In particular 'Seq(' TERM+ ` | ` Var ')'.

ChrisW: Any other comments? Sequence semantics is in the alternative proposal and pairs semantics was used in the original

<Harold> Michael, Pair is a function symbol, so I eliminated that from the syntax, moving it to the semantics.

MichaelK: If you don't have function symbols, you cannot treat it as syntactic sugar

ChrisW: So advantage is you can handle lists without requiring functions

GaryHallmark: It is good to decouple them (lists and function symbols) for production systems

<Hassan> fine

<Hassan> ???

<ChrisW> PROPOSED: Approve Michael's alternative proposal on lists and update FLD+BLD syntax/semantics accordingly to reflect that and the previous resolution on lists

ChrisW: Vote on the above proposal

<sandro> +1

<DougL> +1

<Hassan> 0

<Harold> +1

<IgorMozetic> +1

<LeoraMorgenstern> +1

<MichaelKifer> +1

<mdean> +1

<sandro> Gary on phone: +1

<josb> +1

<ChrisW> RESOLVED: Approve Michael's alternative proposal on lists and update FLD+BLD syntax/semantics accordingly to reflect that and the previous resolution on lists

BLD

ChrisW: Harold, can you give an update on the discussion about presenation grammars for FLD and BLD?

Harold: We agreed at a previous meeting to remove reification from BLD (Basic Logic Dialect)
... we also discussed at F2F10 about making a distinction in the grammar between functions and predicates
... and also to bring in syntax for builtin functions and predicates

ChrisW: And also Jos had an action to add metadata and IRIs to the syntax
... people have agreed to remove reification and to add metadata and IRIs
... so the remaining issue is whether to distinguish between functions and predicates in the grammar

Harold: Michaelk said it is a good idea to keep Uniterm

ChrisW: We are not proposing to remove Uniterms...just in how they are used in the grammar
... yes, it changes the markup by distinguishing functions from predicates
... but still they will have the same syntax

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

Harold: We want to handle future HiLog extensions

ChrisW: Michaelk, where do you stand on this issue? does distinguishing functions and predicates in the syntax make it more difficult to do HiLog extensions?

MichaelK: No, I don't think it does
... that's why I wanted to make the BLD grammar a specialization of the FLD grammar
... so that BLD grammar can be extended in a compatible way

Harold: I'm not convinced this will work. Yes, HiLog would be generalization of BLD

JosB: I proposed 2 grammars: FLD and BLD. the FLD grammar encompasses HiLog

<josb> I give up....

<sandro> josb, is your BLD grammar a subset of your FLD grammar?

<josb> Yes

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

Christian: I don't understand the current discussion. FLD and BLD are the same in the area of predicates and functions

Sandro: I think Harold is saying that if you split uniterm into functions and predicates in FLD then you can't extend to HiLog

<josb> I showed that you CAN!

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

<josb> right

<Harold> We want to read BLD documents (with BLD facts and rules) into future HLD (HiLog) documents.

<Harold> Therefore BLD documents should not separate preds and funcs.

<josb> Harold, just read the grammars I proposed...................

Christian: But HiLog distinguishes between predicates and functions

ChrisW: MichaelK, you made a proposal for the grammars for FLD and BLD.Can you summarize?

MichaelK: I proposed a framework to use around the grammars that JosB had proposed

Harold: I explained my point above in the irc

MichaelK: Harold, I understand that you are saying we need to also consider how it will look in XML, and not just in BNF
... I think it would be possible to accomplish the extensible design in XML
... I wanted to show the concept in BNF, but intended that it would carry over to XML
... However, I didn't think hard about this yet, so can't say for sure whether it is possible

ChrisW: This seems like it should carry over OK to XML

MichaelK: Yes, but it has to be checked

ChrisW: How will we go about checking this?

ChrisW: Someone has to demonstrate that there is an XML syntax that can be specialized from HiLog to BLD

Sandro: Jos says he has done this

MichaelK: Jos hasn't done it for HiLog yet, so he would have to do that

<josb> FLD subsumes hilog

<josb> so, I did it for hilog

<Harold> E.g., the BLD XML-like Atom(a Fun(f c d) e) CANNOT be imported unchanged into HLD.

<Harold> E.g., the BLD XML-like Uniterm(a Uniterm(f c d) e) CAN be imported unchanged into HLD.

Sandro: Why can it not be imported?

<csma> Fallbacks!

ChrisW: Harold, RIF is an interchange syntax. We would not break HiLog by requiring they use this format
...Hilog requires functions to be allowed in places where they are not conventionally used in other languages
... it doesn't require that you don't distinguish between them

<Harold> And ( ?x = Uniterm(f c d) Uniterm(a ?x e) )

<Harold> And ( ?x = Uniterm(f c d) ?x(a ?x e) )

Harold: In above example, ?x occurs in 2 places... at the top level it is an atom
... the other occurrence is not

ChrisW: The distinction is there in what you typed, Why is it a problem to call it out syntactically?

<Harold> At the time you write ?x = Uniterm(f c d) you don't need to say how it's going to be used: So both ?x occurrences in ?x(a ?x e) are fine.

Christian: Problem may occur when using a BLD document in HiLog dialect

<sandro> Sandro: when you parse Harold's expression, you find some occurances of ?x occur in the place where you expect a predicate and some where you expect a function. All I want is the XML grammar to contain those labels from the parsing -- so the parsing work is in the XML, as it's supposed to be with XML.

<josb> right

<josb> +1 to Sandro

Christian: I agree with what Sandro said

<Harold> And ( ?x = Uniterm(f c d) Pre(?x)(a Fun(?x) e) )

Harold: Is the above what you mean, MichaelK?

MichaelK: No

<Harold> And ( ?x = Uniterm(f c d) Pred(?x)(a Fun(?x) e) )

MichaelK: I am not proposing to mark it up. The basic difference between your grammar and JosB's is just at the top level

<Harold> And ( ?x = Uniterm(f c d) ?x(a ?x e) ?x )

Harold: What about the above? is this possible?

MichaelK: Yes, the x's will be marked as atom, but inside they will all be uniterms

ChrisW: Let's move this discussion to email

<sandro> ACTION: Harold to make the case, in e-mail, based on examples in 11 March meeting, for keeping Uniterm in the XML [recorded in http://www.w3.org/2008/03/11-rif-minutes.html#action04]

<trackbot-ng> Created ACTION-445 - Make the case, in e-mail, based on examples in 11 March meeting, for keeping Uniterm in the XML [on Harold Boley - due 2008-03-18].

Christian: We didn't discuss the orthogonal item of having the syntax (presentation and XML) distinguish between logical and builtin functions and predicates

Sandro: We decided on that already

Christian: One grammar proposal distinguishes builtins from logical, and one distinguishes functions from predicates, but neither does both

<Harold> For reference, I talked about Hterms (Uniterm) in the W3C Submission of SWSL-Rules: http://www.w3.org/Submission/SWSF-SWSL/#ruleml-hilog

<Harold> For instance, the HiLog term ?Z(?X,a)(b,?X(?Y)(d)) is serialized as shown below:

JosB: It is still not clear how the XML syntax will be defined i.e. how it relates to presenation syntax

ChrisW: We agreed that the mapping would be in a table, but that the XML syntax would be as close as possible to the presentation syntax so that the mapping will be trivial

Christian: For the predicate production you would need to have 2 entries in the table

JosB: The table is to translate the syntax, it does not care about BNF or schema, just about syntax

JosB: I need to see how the XML can be derived from the BNF - I am skeptical

Hassan: I think it can be derived; I have been working on a tool that can do this

Christian: If we allow metadata inside uniterms for roundtripping purposes...

Hassan: then you need to annotate the BNF

<sandro> hak: you want a forgetful homomorphism

Christian: We may want to have things in the XML syntax that we don't have to reflect in the presenation syntax

ChrisW: Christian, please put your point in an email, with an example

<csma> ACTION: christian to write an example of XML that should not be derived from the BNF of the prez syntax [recorded in http://www.w3.org/2008/03/11-rif-minutes.html#action06]

<trackbot-ng> Created ACTION-446 - Write an example of XML that should not be derived from the BNF of the prez syntax [on Christian de Sainte Marie - due 2008-03-18].

ChrisW: I don't think we should publish the next Working Draft without having the syntactic issues revolved
...We can dedicate next week's telecon to the outstanding syntactic issues

Sandro: And I have two syntactic issues that I will describe in email

Publication plan

ChrisW: Are FLD and BLD ready to be reviewed?

MichaelK: There are some outstanding issues, I sent an email about it
....http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/0052.html
... I will not be at next week's telecon
... I'll plan to make all my changes by Saturday

ChrisW: I think we need to postpone our schedule by one week
... and then reevaluate where we are with syntactic issues
... The actions that were assigned today are critical so that we can resolve syntactic issues at next week's telecon

<csma> +1 to postpone

Christian: Can we talk about JosB's issue about appendix?

<josb> http://www.w3.org/2005/rules/wiki/SWC

JosB: In the current SWC document, the appendix describes embedding, but this is really more of an implementation hint
... so it shouldn't really be part of SWC document, so I'd like to move it to another document that can be published as a Working Group note

ChrisW: You don't like it in the appendix because it makes the document longer?

JosB: No, because it doesn't belong there, because it's a different topic from the main document

<Harold> Jos, Sandro, I think a Working Note is too level a document to be referred to from a Proposed Recommendation.

Sandro: I think people would want it in the same document...it is OK to have-non normative parts of the document

ChrisW: Agree

<IgorMozetic> I'm in favor in keeping it in

JosB: I don't object to leaving it as a non-normative appendix

MichaelK: I don't object either

JosB: Ok, agreed

AOB

adjourned

- DRAFT -

RIF Telecon 11 March 2008

11 March 2008

Attendees

Contents