RIF Telecon 17 March 2009

17 Mar 2009


See also: IRC log


Sandro, csma, DaveReynolds, AdrianP, cke, Gary, [NRCC], +1.631.833.aaaa, Michael_Kifer, Leora_Morgenstern
JosDeBruijn, AxelPolleres
Christian de Sainte Marie
Adrian Paschke



*PROPOSED:* accept minutes of telecon March 3 [1]

<csma> PROPOSED: to approve the minutes of March 3 telecon

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2009Mar/0035.html

<csma> RESOLVED: to approve the minutes of March 3

<csma> PROPOSED: to approve the minutes of March 10

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2009Mar/att-0051/2009-03-10-rif-minutes.html

<csma> RESOLVED: to approve the minutes of March 10

<csma> next item


<csma> next item

Sandro: datatypes in OWL RL
... we need to decide if we object on any of these datatypes
... go for joint value spaces
... list datatypes in OWL RL

Adrian: HCLS looks into query federation for the distributed HCLS KB in Deri and FU Berlin

Adrian: + linked open data interfaces ontop of the KBs


<csma> next item

Actions review

Action 711 closed

<trackbot> Sorry, couldn't find user - 711

action 708 continued

<trackbot> Sorry, couldn't find user - 708

action 701 continued

<trackbot> Sorry, couldn't find user - 701

<scribe> continued action 692

cke: waiting on feedback from Harold

Harold: should have something for the next telecon

cmsa: think there is a test case on UCR 4.1 proposed by Stella

<csma> next item


<csma> 1. Make it clear this is only shorthand for the purposes of writing DTB, and

<csma> that all rulesets must use a fixed arity function/predicate

<csma> 1a. specify one rather than n different functions. So, in the case of concat we

<csma> would only have a binary string concatenation function. Clearly all the others

<csma> can be built from this base case.

<csma> 3a. Remove all the well-formedness requirements. The same symbol can have

<csma> several arities, can be a pred, func, and an individual in different contexts.

<csma> 3b. To keep the separation between preds, funcs, and individuals, but pred,

<csma> func, external symbols can have multiple arities.

<csma> 3c. To keep things as before, but for external symbols to allow multiple arities

<csma> (and maybe even allow them to be funcs and preds in different contexts).

csma: the different options we have

<sandro> -0 option 1 (don't much like it, but it's okay)

<Harold> 0

<DaveReynolds> 0

<Gary> -0


<Michael_Kifer> -1 opt 1

<cke> 0


<DaveReynolds> 0 option 1a

<Gary> +0.5

<sandro> +0 option 1a (only have binary strcat), fixed arity.

<Harold> -0

<Michael_Kifer> -1, I dont understand 1a

<Michael_Kifer> -0.5

<sandro> (mk changes after verbal explanation)

<Gary> +1 for 3a

<sandro> -1 option 3a (I don't think we can get it to work right -- +1 if we could actually implement it)

-1 option 3a

<sandro> jos: -1 option 3a (from e-mail)

<DaveReynolds> +0.5 option 3a if it can be done

<Harold> -0.1

csma: option 3b

<sandro> +1 option 3b

<DaveReynolds> +1 option 3b

<Michael_Kifer> +1, 3b

<sandro> axel: +1 option 3b (from e-mail)

<Gary> +1 for 3b

cmsa: Axel prefers 3b

+1, 3b

<Harold> 0

cmsa: Jos prefers 1 or 1a

option 3c

<sandro> +0.5 option 3c

<DaveReynolds> +0.5 option 3c

<Michael_Kifer> +0.3, 3c

0, option 3c

<Harold> +0.75

<Gary> +0, 3c

<csma> PROPOSED: To keep the separation between preds, funcs, and individuals, but pred, func, external symbols can have multiple arities. Closing ISSUE-92.

<sandro> ciao DaveReynolds

<sandro> csma: do this proposal next week

<csma> next item


csma: proposed solutions will be resolved next week
... bounded quantifiers

<csma> http://www.w3.org/2005/rules/wg/meeting/2009-01-14

csma: reason for bounded quantifiers in PRD
... most PRD engines have them for the reason of efficiency
... PRR has them

<Gary> its trivial syntactic sugar, how can it affect performance?

cmsa: will be needed for an else part

<Gary> its an annoying difference between PRD and Core -- should be in both or better, in neither

<csma> forall ?x, if cond(?x) then action1(?x) else action2(?x)

<Gary> just use 2 rules

<sandro2> testing.

<adrianpaschke> just dropped out of irc

<Gary> not need in Oracle Business Rules, maybe ILOG

<Gary> not a great reason to add to PRD either, IMHO

<adrianpaschke> csma: logic rules do not have an else part?

<adrianpaschke> csma: what would be a reason to add else in Core?

<adrianpaschke> Sandro: convient write rules as if-then-else rules

<sandro> not "else" -- just Bounded Quantifiers are nice.

<Gary> you really want people trying to "performance tune" their rules by moving conditions to different parts of a rule?

<adrianpaschke> Sandro: write rules with bounded quantifiers - readability issue

<Gary> Christian mentioned performance

<adrianpaschke> Sandro: for me the reason is just readability

<cke> this is a semantic issue: bounded quantifier qualify the objects for if / else rules

<adrianpaschke> csma: perfomance was the reason for PRD

<Michael_Kifer> I don't think this is a common issue

<adrianpaschke> csma: bounded quantifier - would they effect the safeness condition

<Harold> Bounded quantifiers remind me of sorted variables, which came into BLD, then were taken out...

<adrianpaschke> csma: if bounded quantifier tells which are all the values you can bind -> you are safe

<Gary> I think no impact on safeness

<adrianpaschke> Harold: bounded quantifiers are not in BLD

<adrianpaschke> Harold: similar to sorted variables, which were taken out

<adrianpaschke> Harold: not in last call BLD

<adrianpaschke> michael: I don't we can add them to Core without changing BLD

<adrianpaschke> csma: just try to understand the impact - if there is a reason for a second last call

<adrianpaschke> michael: impact is - redo BLD

<adrianpaschke> Sandro: for BLD it is just syntactic sugar

<adrianpaschke> michael: need to check consistency

<adrianpaschke> Sandro: not a priority issue - just if we have time

<adrianpaschke> michael: in general bounded quantifiers are useful

<adrianpaschke> michael: will make it easier to express things

<adrianpaschke> michael: optimization

<Gary> remove from PRD: +1, add to Core: +0.5, leave in PRD but nowhere else: -0.5

<cke> can we keep bounded quantifier in PRD, even it is not in Core?

<adrianpaschke> michael: it is syntactic sugar that helps

<adrianpaschke> csma: wondering if in other logical dialects bounded quantifiers are more than syntacty sugar

<adrianpaschke> michael: onyl few languages have them

<adrianpaschke> miachel: e.g. mecury

<adrianpaschke> csma: not so many benefits to add them to Core

<adrianpaschke> csma: not priority to add them in Core

<sandro> csma: I'm hearing not much enthusiasm for them in Core.

<adrianpaschke> michael: in the body of a rule such a quantifier is not very useful

<adrianpaschke> michael: useful for universal quantifiers

<Harold> Couldn't bounded quantifiers become another syntactic desugaring effort on top of RIF, rather than part of RIF?

<sandro> PROPOSED: (Only if we're doing a second last call of BLD anyway) Add Bounded Quantifiers to Core and BLD.

<cke> An example of rule with bounded quantifier: (forall customers such that age > 18 and city is Paris) (if the customer is rich then do something else do something). The first part is a bounded quantifier, it defines the objects for the rule.

<adrianpaschke> michael: univerisal quantifiers in the rules body help simplify the expression

<sandro> +1 (but not enough to slip the schedule more than about two weeks)


<Gary> +0.5 (but rather remove from PRD)

<Harold> -0.5

<Michael_Kifer> -0.1

<adrianpaschke> +1

<adrianpaschke> for translating from PRD into Core a translator would need to create two rules from one with bounded quantifiers

<Michael_Kifer> sandro, I think thee sched will slip > 2 wks because of these quantifiers

<csma> next item


<csma> (2) Leave pred:literal-equal as is

<csma> [on the grounds of symmetry with pred:literal-not-equal, accepting there

<csma> is some redundancy.]

<csma> (1) Drop pred:literal-equal (retaining pred:literal-not-equal)

<csma> [This still leaves me able to shorten the OWL 2 RL and similar rules.]

<csma> (3) Redefine pred:literal-equal to perform all the datatype specific

<csma> equality tests (and redefine pred:literal-not-equal compatibly so that

<csma> for any pair of literals exactly one of these predicates is true).

<csma> [This means I can't use those predicates for the OWL 2 RL rules easily

<csma> (assuming OWL opt for disjoint value spaces). However, there may still

<csma> be value in such predicates for other users.]

<csma> (0) Drop both pred:literal-equal and pred:literal-not-equal

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2009Mar/0076.html

<adrianpaschke> csma: four options

<adrianpaschke> csma: pred_literal-equal(x,y)

<adrianpaschke> csma x and y are literal and x=y

<sandro> sandro: "as is" is: literal-equal(x,y) iff literal(x) and literal(y) and x=y

<sandro> LeoraMorgenstern, you can see the IRC log so far there.

<sandro> strawpoll: option "(2)" pred:literal-equal "as is" --- literal-equal(x,y) iff literal(x) and literal(y) and x=y

<sandro> +0.5

<adrianpaschke> +0.5

<Michael_Kifer> -0.1

<Gary> -0.5

<adrianpaschke> Dave: +1 (according to his email)

<adrianpaschke> Jos: prefers not to have generic predicates and if we have them prefers XPath built-ins (by email)

<adrianpaschke> Jos: -0.5 (according to email)

<sandro> STRAWPOLL: option "(1)" Drop pred:literal-equal (retaining pred:literal-not-equal)

<sandro> +1

<adrianpaschke> -1

<Gary> -0.99

<Michael_Kifer> +0.8

<adrianpaschke> yes, for usability reasons it is good to have them both

<sandro> STRAWPOLL: option "(3)" Redefine pred:literal-equal to perform all the datatype specific ... ("xpath equality")

<Gary> +0.99

<sandro> +0 nice for users, maybe hard to implement

<adrianpaschke> +1

<adrianpaschke> Dave: -0.5. (according to email)

<adrianpaschke> Jos: -0.5 (according to email)

<Michael_Kifer> 0

<adrianpaschke> cke: 0

<sandro> STRAWPOLL: option "(0)" Drop both pred:literal-equal and pred:literal-not-equal

<Gary> +1

<sandro> -0.25

<adrianpaschke> +1

<adrianpaschke> Jos: +1 (according to email)

<sandro> -1

<sandro> (it's about extensibility)

<sandro> -0.9

<sandro> (it's about being able to write an OWL-RL ruleset that doesn't know what datatypes are supported.)

<sandro> Dave: -0.5 (according to e-mail)

<adrianpaschke> csma: probably option 3

<Michael_Kifer> 0

<Gary> theory guys: I guess adding != (i.e. Not(a=b)) breaks everything?

<Gary> seems like we are hung up on owl:different

<adrianpaschke> csma: we have discussed this issue from the OWL-RL point of view

<adrianpaschke> csma: general literal-equal will provide extensibility

<adrianpaschke> Sandro: except of a non-identity test

<adrianpaschke> Sandro: can have all the XPath indentity tests

<sandro> STRAWPOLL: Have xpath-style-equals, xpath-style-not-equals, AND non-identical-literals (eg for OWL RL).

<sandro> "non-identical-literals" == current pred:literal-not-equals

<sandro> +0.5 it'd be nice, but it sounds like too much work.

<cke> +0.5

<Michael_Kifer> 0

<adrianpaschke> +0.5

<Gary> +9

<Gary> oops, +.9

<csma> PROPOSED: Have xpath-style-equals, xpath-style-not-equals, AND pred:literal-not-equal for non-identical-literals (eg for OWL RL). Closing ISSUE-80.

<adrianpaschke> Sandro: worried about the work

<adrianpaschke> * Axel is DTB author ;-)

<csma> next item


<adrianpaschke> csma: important issue for PRD

<adrianpaschke> csma: almost finished by new strawman

<adrianpaschke> csma: and we have Gary starwman

<Gary> -1

<Michael_Kifer> -9

<Michael_Kifer> no, -9

<adrianpaschke> csma: big benefit is simplicity and implementability of the proposal

<adrianpaschke> csma: two cons; XML document without a schema, you need a schema

<adrianpaschke> csma: cannot work with data without a schema

<adrianpaschke> csma: second issue, if we import RDF document as XML

<adrianpaschke> csma: will have a completey different RIF document

<adrianpaschke> csma: than if we use SWC for RDF intergration

<adrianpaschke> csma: my proposal can handel data with and without schema

<adrianpaschke> csma: compatible with SWC in the sense that if you import RDf document as RDF/XML (given a view conventions) you will have the same document

<adrianpaschke> csma: with the same interpretation

<adrianpaschke> csma: contrast to Michael - I think it is quite elegant

<adrianpaschke> csma: but strawman is not finished yet

<adrianpaschke> csma: not sure how easily implementable it is

<adrianpaschke> csma: easy if you know the schema and mapping to your platform-sepcific language

<adrianpaschke> csma: if you need to do the mapping on the fly

<adrianpaschke> csma: proposal orthorgonal to Gary's proposal

<sandro> didn't read it yet, sorry :-(

<adrianpaschke> michael: proposal is a good starting point - send an email about what I don't like

<adrianpaschke> michael: can be probably improved

<adrianpaschke> csma: would like that someone takes a look if it is compatible with SWC


Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.133 (CVS log)
$Date: 2009/03/17 16:32:36 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
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)

Found Scribe: Adrian Paschke
Found ScribeNick: AdrianP
Default Present: Sandro, csma, DaveReynolds, AdrianP, cke, Gary, [NRCC], +1.631.833.aaaa, Michael_Kifer, Leora_Morgenstern
Present: Sandro csma DaveReynolds AdrianP cke Gary [NRCC] +1.631.833.aaaa Michael_Kifer Leora_Morgenstern
Regrets: JosDeBruijn AxelPolleres
Agenda: http://lists.w3.org/Archives/Public/public-rif-wg/2009Mar/0077.html
Got date from IRC log name: 17 Mar 2009
Guessing minutes URL: http://www.w3.org/2009/03/17-rif-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

[End of scribe.perl diagnostic output]