RIF Telecon 23 Oct 2007 -- 23 Oct 2007

<ChrisW> Scribe: Igor

<csma> Meeting: RIF telecon 23 October 2007

<csma> Chair: Christian de Sainte Marie

<csma> Scribe: Igor Mozetic

<csma> topic #rif 23 October RIF telecon agenda: http://lists.w3.org/Archives/Public/public-rif-wg/2007Oct/0097.html

<csma> scribenick: IgorMozetic

Admin

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2007Oct/att-0078/16-Oct-2007-rif-minutes.html

accepted minutes of Oct 16 telecon

Liaison

no actions

F2F8

post regrets if you don't plan to attend

<csma> http://www.w3.org/2005/rules/wg/wiki/F2F8

<JeffP> should be in the minutes of last telecon

<csma> http://lists.w3.org/Archives/Public/public-rif-comments/2007Oct/0000.html

public comments on RIF-BLD WD2 (as of Oct 15) by PFPS

we will address public comments after the publications of WD2 on Oct 31

<sandro> many apologies for being so late.

sandro: no updates regarding F2F8

the publication date, 31 Oct, of RIF-BLD WD2 is tentative

send suggestions for the F2F8 agenda to the rif mailing list

BLD+SWC editors drafts

sandro: 15 and 17 people registered for the two F2F8 meeting days, respectively

next telecon: decide to publish or not the WD2

the intend was to publish BLD+SWC together

csma: we should make clear that SWC is WD1, and we will publish BLD WD2 and SWC WD1 separately

<sandro> csma: we can make the decisions separately

csma: the point in publishing SWC is to make clear that we did work and want to get public comments

<Zakim> sandro, you wanted to ask about shortnames

sandro: we need short names for the documents

<Harold> +1 to Christian

csma: let's call it BLD WD2

sandro: will check if this is possible

<Harold> (as planned, just changing the name but not the versioning numbers)

<josb> for the swc doc: http://www.w3.org/TR/rif-rdf-owl?

harold: there was official decision to split core into two docs

<sandro> http://www.w3.org/News/2007

<sandro> http://www.w3.org/News/2007#item57

harold: proposes the name change Core-> BLD and continue with versioning, i.e., WD2

<sandro> http://www.w3.org/TR/rif-core/

sandro: the above is the first version

<sandro> rif-bld will refer to http://www.w3.org/TR/2007/WD-rif-core-20070330/ as its "Previous Version", formally (if possible) or informally

harold: proposes a master document to point to other docs

csma: BLD WD2 should point back to Core WD1 as previous version

<sandro> rif-swc ?

csma: proposes short names rif-bld and rif-swc

<sandro> rif-rdf-owl

<sandro> rif-rdf-owl-compat

josb: prefers the above over SWC

igor: +1

<sandro> rif-ucr, rif-core, rif-bld, rif-rdf-owl

no objections to the above proposal

<sandro> PROPOSED: Short name request for Jos' document will be rif-rdf-owl

<sandro> RESOLVED: Short name request for Jos' document will be rif-rdf-owl

<sandro> PROPOSED: short name request for BLD will be rif-bld, with rif-core being republished as a trivial document referring people to rif-bld.

no objections to the above proposal

<sandro> Axel: but what if we have an absyn for RIF that is written in RDF and OWL? Wont that name get confusing?

<Harold> I was referring to this farily official looking 'Core splitting' news/blog item, which could become a starting point of a stub document branching to rif-bld and later also to rif-prd: http://www.w3.org/blog/SW/2007/07/20/rif_wg_agrees_to_basic_logic_and_pr_dial

<sandro> Axel: but I can live with it.

<sandro> RESOLVED: short name request for BLD will be rif-bld, with rif-core being republished as a trivial document referring people to rif-bld.

<sandro> Harold, I think the news item you saw was this: http://www.w3.org/blog/SW/2007/07/20/rif_wg_agrees_to_basic_logic_and_pr_dial

Test cases

<csma> http://www.w3.org/TR/owl-test/

<sandro> test cases as : input + output, or input+query+query-results

<sandro> Sandro: I don't think anyone has proposed intermixing print statements into the bodies of rules --- just a special kind of rule with a print body.

josb: rif doesn't define query language

<sandro> Jos: test entailments, but these are not queries.

<sandro> Maybe conclude "output(....)" which can be seen as an action or a conclusion fact.

<josb> sure

<josb> right

<DaveReynolds> yes

<josb> also conditions with existential quantifiers

<sandro> So -- PositiveEntailmentTests and NegativeEntailmentTests ?

<Harold> Jos, DaveR, the Condition Language is mentioned to be usable as a query language. Usually you write a query rule of the form answer(?x1 ... ?xN) :- rdf-query-expression-with-free-variables(?x1 ... ?xN).

<Harold> The N variables of answer(?x1 ... ?xN) will then be printed by the rule engine.

<sandro> We need to clarify the difference between print, query, and entailment testing.

csma: we have the print, query and the entailment approaches

<DaveReynolds> Harold - that is getting into notions of API and query that we don't want to standardize and is not necessary. We can just use the rule language to express the condition and whose conclusion is a simple ground predicate so we don't need to include variable bindings.

<sandro> Jos: what we specify is entailment, so that's what should be checked.

<sandro> PROPOSED: We'll use PositiveEntailmentTests and NegativeEntailmentTests, and we can explain them in terms of Queries and Rrints for the relevant communities.

<sandro> PROPOSED: We'll use PositiveEntailmentTests and NegativeEntailmentTests, and we can explain them in terms of Queries and Prints for the relevant communities.

<sandro> PROPOSED: We'll use PositiveEntailmentTests and NegativeEntailmentTests (along with syntactic tests, and maybe others), and we can explain them in terms of Queries and Prints for the relevant communities.

igor: with entailment we are testing the soundness and completness of inference engines; we need to test the interchange format, ie, a syntax test

josb: agrees with Igor to have also separate syntactic tests

<sandro> Sandro: This document {....} is / is-not a syntactically valid BLD document.

<sandro> Sandro: This is separate from conformance, which speaks to what you actually DO on getting one of those.

<JeffP> Sorry - I have to leave now.

<JeffP> Bye

<ChrisW> igor, a little more detail on scribing would be helpful

<sandro> Sandro: There is a subtle (and perhaps unimportant) difference between saying "X is a syntactically valid BLD document", and "All implementations must correctly report that X is a syntactically valid BLD document"

PROPOSED: also include syntactic tests

<Harold> Sandro, Christian, while validity tests can be used for all kinds of XML-based specs (SVG, SOAP, WSDL, ...), entailment tests can only be used for logic-based languages (RDF, SPARQL, OWL, RIF, ...).

dave: test if a specified predicate is entailed or not

<sandro> PROPOSED: We'll use PositiveEntailmentTests, NegativeEntailmentTests, PositiveSyntaxTests, NegativeSyntaxTests, (and maybe others). We'll explain that entailment tests can be implemented by rules which conclude a ground fact which you check for.

<Harold> ... only after validity is ensured, can you move on to entailment tests.

<sandro> Dave: I was understanding the ground-fact would be in the test case

proposal: use only ground queries, and the entailement test is yes/no

csma: expect the output to list all the entailed ground facts

<Harold> Christian, "all the ground facts" works only for finite models.

<sandro> csma: A BLD input document, and an output document which contains a set of all the ground facts which are entailed.

<Harold> ... so only for the Datalog subset of RIF.

<ChrisW> lost the scribe

<sandro> sandro: should not be ALL entailments -- but a specific one.

<sandro> sandro: so you only want to be able to test ground entailments???

josb: we should test all closed condition formulas

<sandro> Jos: I think we should test what the def'n allows us to test.

<sandro> Jos: The defn of entailment defines when a particular BLD ruleset entails a closed bld condition formula.

<sandro> Jos: So an entailment test case has the same elements.

josb: we should have also exist. quantified formulas to test

<sandro> Dave: could be --- that closed formula is the body of another rule.

<sandro> Jos: I agree that they are equivalent.

josb: if we make the equivalence clear in the test cases document, jos agrees with Dave's proposal

<sandro> Sandro: My inclination is to keep that extra rule out of the entailment tests.

<Zakim> GaryHallmark, you wanted to ask why we are avoiding just adding queries

gary: why not include queries?

csma: doesn't agree to include queries in RIF

gary: seems simple, and one can write a portable test case

<Harold> For getting a Query Language from the Condition Language, the found bindings of the top-level Exists in Condition Language formulas need be just be printed.

sandro: there is no portable way to talk to any RIF system (no API)

<Harold> Gary, doing printing only on the top-level is no problem, not even semantically (also see 'monads' in functional programming).

<GaryHallmark> if you have a query that returns a RIF document containing ground facts, then you have a portable test

sandro: create a wiki page with test cases proposal, examples

<StellaMitchell> I could try

<scribe> ACTION: stella will create the wiki page on test cases [recorded in http://www.w3.org/2007/10/23-rif-minutes.html#action01]

<rifbot> Created ACTION-360 - Will create the wiki page on test cases [on Stella Mitchell - due 2007-10-30].

<sandro> ACTION: StellaMitchell to update the Test Cases wiki page to give us a more concrete proposal (or proposals) with one or more examples. [recorded in http://www.w3.org/2007/10/23-rif-minutes.html#action02]

<sandro> ACTION: Stella to update the Test Cases wiki page to give us a more concrete proposal (or proposals) with one or more examples. [recorded in http://www.w3.org/2007/10/23-rif-minutes.html#action03]

<rifbot> Created ACTION-361 - Update the Test Cases wiki page to give us a more concrete proposal (or proposals) with one or more examples. [on Stella Mitchell - due 2007-10-30].

<StellaMitchell> yes

to scribe next week: Leora

Extensibility issues

<sandro> http://www.w3.org/2005/rules/wg/wiki/Arch/Extensibility_Design_Choices

sandro: have a look at the above page and add comment, pros/cons, ...

<Harold> Also see http://www.w3.org/2005/rules/wg/wiki/Arch/RIF_Components/RIF_Dialect_Structure for a system of such extensions (Naf etc.): http://www.w3.org/2005/rules/wg/wiki/Arch/RIF_Components/RIF_Dialect_Structure

csma: next week telecom in Europe at 16:00 (instead of 17:00)

adjourned

- DRAFT -

RIF Telecon 23 Oct 2007

23 Oct 2007

Attendees

Contents