W3C

- DRAFT -

RIF Telecon 9-Sep-2008

09 Sep 2008

Agenda

See also: IRC log

Attendees

Present
ChrisW, csma, Harold, Sandro, LeoraMorgenstern, Dave_Reynolds, Hassan_Ait-Kaci, ChanghaiKe, GaryHallmark, StellaMitchell, AdrianPa, Michael_Kifer
Regrets
JosDeBruijn, AxelPolleres, StuartTaylor
Chair
Chris Welty
Scribe
DaveReynolds

Contents


Admin

<ChrisW> http://lists.w3.org/Archives/Public/public-rif-wg/2008Sep/att-0043/20080902-rif-minutes.html

<ChrisW> PROPOSED: accept minutes from Sept 2 telecon

<ChrisW> RESOLVED: accept minutes from Sept 2 telecon

Liason

Sandro: OWL WG rule profile is progressing, at some point should look at the rules and check can be written in BLD
... compare with our embedding, etc.
... Jos critical path for some of that?

<Harold> 4.3 Reasoning in OWL 2 RL and RDF Graphs using Rules

<Harold> http://www.w3.org/2007/OWL/wiki/Profiles#Reasoning_in_OWL_2_RL_and_RDF_Graphs_using_Rules

ChrisW: Impact on last call?
... there were OWL WG participants in the SWC document, does OWL-2 have any impact on that?

Sandro: no evidence there is a problem, but should have someone who knows what they are doing look it over.

<ChrisW> ACTION: sandro to ask for a review of RDF&OWL from OWL-wg [recorded in http://www.w3.org/2008/09/09-rif-minutes.html#action01]

<trackbot> Created ACTION-573 - Ask for a review of RDF&OWL from OWL-wg [on Sandro Hawke - due 2008-09-16].

<ChrisW>

<ChrisW> ACTION: Dave to look at OWL-2-RL rules and consider whether they are implementable in BLD [recorded in http://www.w3.org/2008/09/09-rif-minutes.html#action03]

<trackbot> Created ACTION-574 - Look at OWL-2-RL rules and consider whether they are implementable in BLD [on Dave Reynolds - due 2008-09-16].

Publicity

ChrisW: there is a Wiki page for linking external announcement postings. Check that publicity actions have been done but also recorded there.

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

<sandro> Harold: Michael posted to xsb, flora-users, [etc]

csma: have new public comment

<csma> http://lists.w3.org/Archives/Public/public-rif-comments/2008Sep/0000.html

ChrisW: need to start a wiki page on it, it is mostly about FLD

Harold: Michael is aware of it

Harold: created a page for response to Richard O'Keefe comments

<Harold> http://www.w3.org/2005/rules/wiki/Response_to_OK

Action review

ACTION-572 completed by Jos

ACTION-570 pending discussion

<AdrianPa> yes, will put it to top priority

<Hassan> yes

ACTION-564 continued, waiting on APS discussions

Hassan: waiting for syntax to settle

F2F11

<sandro> http://www.w3.org/2002/09/wbs/38457/f2f11reg/

ChrisW: F2F11 proceeding as planned. Any objections to the proposed dinner should be raised now.

Core

<csma> PROPOSED: Core will not have named-argument uniterms

ChrisW: seemed to be consensus on closing this at last discussion.

Gary: so why are they in BLD?

ChrisW: it was a close decision even there.
... at that point there was an explicit flag that being in BLD didn't mean necessarily having them in Core

Gary: this doesn't seem like a painful thing to implement

Sandro: don't object but it would be nice to have a clear rationale to explain the decision to others

csma: want to keep the option of dialects that don't have it, if it is in Core then all dialects have to have it

<sandro> csma: reason -- we want to keep the possibility of having dialects that don't have it. that's why not in core.

<scribe-note> RESOLVED: Core will not have named-argument uniterms

ChrisW: current open issues for Core - classification constructs (46), decideable (71), access to external functions (71), skolem functions (72), no predicates (74), disjunction (75), equality (76)

Harold/Gary: disjunction not a no-brainer. Gary would like it or some equivalent.

Harold: equality not-in-head already agreed, 76 is about equality in body

ChrisW: straw poll for NOT equality-in-head in Core

[No objections voiced]

ChrisW: what would be needed to resolve 76?

Harold: relates to the issue of builtins - predicates or functions

<AdrianPa> we discussed restricted equality in the body

Michael: disagrees, just an identity relationship, sees no reason not to include it

Hassan: becomes syntactic equality, most trivial

<Harold> Even if you don't have an equality construct for the body, everyone could define it in one fact: eq(?x ?x).

<AdrianPa> in PRD it is a single assignment function

<AdrianPa> slightly different from equality assignment which might reduce to identy equality if both sides are bound

<Hassan> = then means are these two pointers identical

<Hassan> Can we have an example of where it is useful?

Dave: doesn't this require equation rewriting

Gary: yes but need that any way

<Harold> http://www.w3.org/2005/rules/wiki/DTB#pred:numeric-equal_.28adapted_from_op:numeric-equal.29

<csma> Adrian, what do you mean with "in PRD, it is a single assignment function"? You mean: equality?

<GaryHallmark> A(?x) :- A(?x - 1) requires expression rewriting for PRD

<Hassan> I agree with Dave. We should spell out the criteria of what is or isn't in Core.

<sandro> I think so, yes, some builting in core

<GaryHallmark> i.e. rewrite to A(?x+1) :- A(?x)

<Hassan> Arithmetic?

Michael: doesn't require any new mechanism beyond what you need for builtins anyway

Dave: but for builtins we have on the table the possibility of binding patterns to restrict use, that wouldn't apply to syntax constructs like this equality

Michael: binding pattern's aren't going to fly in Core anyway

Sandro: trying to see where someone would need it

<Harold> My above eq(?x ?x) could be used for (single-)assignments, the above pred:numeric-equal(?arg1 ?arg2) could not.

Michael: consider use in prolog

<Harold> This is because of the modes ("binding patterns")

<Harold> eq(?myvar 1) is fine.

Dave: what about SWC and interaction with owl:sameAs?

Michael: it is not identity over datatypes

<Harold> pred:numeric-equal(?myvar 1) is not fine.

Dave: is there a difference between this equality and an untyped equality buitin?

ChrisW: seems the same

ChrisW: wants to get this one closed, at least agree on a resolution next week

PRD

<Hassan> gotta go - bye

ChrisW: This is about action-554 and issue-66

Gary: made several suggestions on object creation
... basic issue is whether there is something that should be shared with Core/BLD

<AdrianPa> I would vote for a special new built-in, too

<csma> Why is that a builtin?

<AdrianPa> this would allow new logic RIF dialects which support object creation

<csma> I mean, why is that "New" construct a builtin, not an action (like Assert, Retract, etc)?

Gary: proposed options including builtin, skolem function like approach and new syntax
... so would the group like something that can be in Core or should PRD go off in its own direction?

csma: don't understand why it is sometimes referred to as a "builtin", isn't it another action alongside assert?
... consider RETRACT, there is an action and associated keyword (same for ASSERT even if done differently)
... NEW is not a builtin because if you call it multiple times it gives different instances

Gary: yes, hence the need for an occurrence number - confusing the different options

Adrian: advantage of a builtin is that it can be used in other dialects, so can call it in the body of a rule

Gary: builtin is a pure mathematical function, fixed interpretation function, so Gensym is not legal

csma: preference is for option (a) then?

<AdrianPa> Stakeholder for Assert and Retract are is e.g. a RIF Prolog dialect

<GaryHallmark> http://lists.w3.org/Archives/Public/public-rif-wg/2008Sep/0087.html

<AdrianPa> Most prolog engines support assert and retract built-ins in the body of a rule

Dave: do have some interest in option C, relates to issue of Skolem functions in Core.

ChrisW: so if have (a) in PRD and Skolem in Core how would they interact?

Gary: the import becomes more difficult because trasnlator might be faced with either formulation and have to convert between the two.

csma: Object creation is intrinsically non-logical concept, if we used skolem functions for this doesn't this confuse things at least for implementers?

ChrisW: need to get these issues moved to closure, time is running out
... would like Gary and Harold to organize telecons for PRD and Core to progress these issues further

<AdrianPa> yes, good idea. Let's have a PRD task force with telecons

csma: include others, not just editors, not closed

ChrisW: yes, announce telecons on the mail list so stakeholders can participate

csma: resolutions would still be made in the whole WG telecons

ChrisW: free to use Zakim bridge for this

Test Cases

<AdrianPa> http://www.w3.org/2005/rules/wiki/Category:Test_Case

ChrisW: plan is to go through TestCases and approve them
... in first OWL WG used these to give evidence that there are implementations and they are doing what the spec says

<AdrianPa> unfortunately I will be travelling next week

<ChrisW> zakim: take up item 10

<AdrianPa> so can not particiapte but will cordinate with the other test case editors

<AdrianPa> bye

<ChrisW> scribe for next week: MichaelKifer

Summary of Action Items

[NEW] ACTION: Dave to look at OWL-2-RL rules and consider whether they are implementable in BLD [recorded in http://www.w3.org/2008/09/09-rif-minutes.html#action03]
[NEW] ACTION: DaveReynolds to look at OWL-2-RL rules and consider whether they are implementable in BLD [recorded in http://www.w3.org/2008/09/09-rif-minutes.html#action02]
[NEW] ACTION: sandro to ask for a review of RDF&OWL from OWL-wg [recorded in http://www.w3.org/2008/09/09-rif-minutes.html#action01]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.133 (CVS log)
$Date: 2008/09/09 16:31:08 $