SV_MEETING_TITLE -- 26 Sep 2008

<AxelPolleres> scribe: AxelPolleres

<scribe> scribenick: AxelPolleres

chris showing the meeting objectives, cf. http://www.w3.org/2005/rules/wiki/F2F11

chris: we must progress on the test cases before going to CR.
... main issue.
... agenda, cf. http://www.w3.org/2005/rules/wiki/F2F11

sandro: how shall we change morning agenda, since we lost already 1/2 hr.

adrian: we can cut tc to 1/2 hr

<AdrianP> RIF test cases

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

Test Cases

csma: let's go through them one by one.

<sandro> subtopic: http://www.w3.org/2005/rules/wiki/Annotation_Entailment

1) http://www.w3.org/2005/rules/wiki/Annotation_Entailment - by jos.

scribe: found an error.

chris: let's fix simple errors on the fly.

jos: testing whether there is an annotation in the ontology imported.

sandro: how do you know that it is an annotation property.

jos: for all objects of type owl:Ontology, all properties must be annotation properties...
... will check again.

dave: not sure, whether dc:title should be declared an annotation property.

<sandro> add dc:title rdf:type owl:AnnotationProperty

axel: I think hijacking dc:title that way is VERY weird.

sandro: right, you can't use dc:title then in an ontology you merge.

axel: you can just create a subproperty of dc:title.

christian: next test case.

<sandro> subtopic: AnnotationPropertyID rdf:type owl:AnnotationProperty

<sandro> subtopic: http://www.w3.org/2005/rules/wiki/Arbitrary_Entailment

http://www.w3.org/2005/rules/wiki/Arbitrary_Entailment

jos: that is the simplest form of inconsistency in RIF.

adrian: should we define a separate test category for testing inconsistency?

agreement on that trst case.

chris: a birt more description would be in order.
... resolve this status to "approved"?

PROPOSED: Accept http://www.w3.org/2005/rules/wiki/Arbitrary_Entailment

<sandro> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Arbitrary_Entailment

<sandro> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Arbitrary_Entailment with more detailed description

<sandro> ACTION: jos add more description to http://www.w3.org/2005/rules/wiki/Arbitrary_Entailment [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action01]

<trackbot> Sorry, amibiguous username (more than one match) - jos

<trackbot> Try using a different identifier, such as family name or username (eg. jdebruij2, jderoo)

<sandro> ACTION: jdebruij2 add more description to http://www.w3.org/2005/rules/wiki/Arbitrary_Entailment [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action02]

<trackbot> Created ACTION-580 - Add more description to http://www.w3.org/2005/rules/wiki/Arbitrary_Entailment [on Jos de Bruijn - due 2008-10-03].

<sandro> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Arbitrary_Entailment with more detailed description

http://www.w3.org/2005/rules/wiki/Chaining_strategy_numeric-add_1

by Axel

<sandro> Jos: BLD 2.3 says your can't do this -- no externals in head

<sandro> Jos: oh... no, that's just atoms

<sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/Chaining_strategy_numeric-add_1

<sandro> RESOLVED: Approve http://www.w3.org/2005/rules/wiki/Chaining_strategy_numeric-add_1

<sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/Chaining_strategy_numeric-add_2

<sandro> RESOLVED: Approve http://www.w3.org/2005/rules/wiki/Chaining_strategy_numeric-add_2

sandro: the description of subtract-variants should reference the addition variants, highlighting the duality between them.

axel: I should add a variation testing for a(11) without the "termination condition" X > 0.

<sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/Chaining_strategy_numeric-subtract_1

<sandro> ACTION: Axel improve the description of chaining strategies [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action03]

<trackbot> Created ACTION-581 - Improve the description of chaining strategies [on Axel Polleres - due 2008-10-03].

<sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/Chaining_strategy_numeric-subtract_1 and 2

<sandro> RESOLVED: Approve http://www.w3.org/2005/rules/wiki/Chaining_strategy_numeric-subtract_1 and 2

http://www.w3.org/2005/rules/wiki/Class_Membership

by adrian.

jos: this is not correct.
... membership can't be used as terms.

christian: adrian, please adapt, then we can revisit that test case.

http://www.w3.org/2005/rules/wiki/Classification_non-inheritance

by stella.

dave: This testcase shows that RIF doesn't have something like class properties.

<sandro> Chris: Instances do not inherit the properties of their classes [ Classification non-inheritance]

<sandro> Chris: You use formulas, not facts, to give properties of all instances of a class.

<sandro> Chris: It looks kind of like a default, but BLD doesn't have anything like that.

Dave: if you approach this from an OOP viewpoint, you could expect the conclusion.

<sandro> Dave: Someone coming at frames from Java might try to do this kind of (disallowed) inheritance

Dave: that is what that TC shows.

Axel: we could add a rule in a variation that does entail the conclusion.

<sandro> PROPOSED: Accept http://www.w3.org/2005/rules/wiki/Classification_non-inheritance

<sandro> csma: instead of using favoritePerson, let's use cardinality->6 billion

christian: the description should be extended to discuss properties of classes and instances.

Leora: Would it be helpful to have a variant as suggested by axel?

<sandro> ACTION: axel add test case related to http://www.w3.org/2005/rules/wiki/Classification_non-inheritance that makes all class properties apply to instances [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action04]

<trackbot> Created ACTION-582 - Add test case related to http://www.w3.org/2005/rules/wiki/Classification_non-inheritance that makes all class properties apply to instances [on Axel Polleres - due 2008-10-03].

?X[?P -> ?V] :- And ( ?X#?Y ?Y[?P -> ?V] )

hi back!

:-)

variation of http://www.w3.org/2005/rules/wiki/Classification_non-inheritance accepted.

<sandro> RESOLVED: Accept http://www.w3.org/2005/rules/wiki/Classification_non-inheritance

http://www.w3.org/2005/rules/wiki/Disjunctive_Information

by jos.

jos: that's a nasty one, it needs a constraint solver.

has equality in the head, which is at risk.

Sandro: description should include evil grin of jos :-)

jos: without negative guards we can't have disjunction here.

chris: as long as we have equality in the head, it makes sense to have that.

sandro: a little bit more description is in order.

<sandro> RESOLVED: Accept http://www.w3.org/2005/rules/wiki/Disjunctive_Information

http://www.w3.org/2005/rules/wiki/EntailEverything_1

<sandro> ACTION: jos To explain better how/why http://www.w3.org/2005/rules/wiki/Disjunctive_Information is so tricky [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action05]

<trackbot> Sorry, amibiguous username (more than one match) - jos

<trackbot> Try using a different identifier, such as family name or username (eg. jdebruij2, jderoo)

by gary

<sandro> ACTION: jdebruij2 To explain better how/why http://www.w3.org/2005/rules/wiki/Disjunctive_Information is so tricky [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action06]

<trackbot> Created ACTION-583 - Explain better how/why http://www.w3.org/2005/rules/wiki/Disjunctive_Information is so tricky [on Jos de Bruijn - due 2008-10-03].

gary: this is waht PR systems typically can't do.
... http://www.w3.org/2005/rules/wiki/EntailEverything_1 - http://www.w3.org/2005/rules/wiki/EntailEverything_6 are basically all the same.

<sandro> ACTION: josb remind people that "josb" works for assigning actions to him. [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action07]

<trackbot> Created ACTION-584 - Remind people that \"josb\" works for assigning actions to him. [on Jos de Bruijn - due 2008-10-03].

Axel: http://www.w3.org/2005/rules/wiki/EntailEverything_1 doesn't need the p(1 2) fact.

<sandro> ACTION-584 done

<sandro> ACTION-584 closed

<trackbot> ACTION-584 Remind people that "josb" works for assigning actions to him. closed

christian/jos/gary desicuss that we could unify all those into one test case.

scribe: with a conjunction in the conclusion.

christian: can we make the decision right now?

http://www.w3.org/2005/rules/wiki/Equality_in_conclusion_1

by adrian

adrian: that's an implementation of factorial function

harold: And( .... ) is missing in the body, forall missing, infix-arithmetics not allowed.

<sandro> subtopic: http://www.w3.org/2005/rules/wiki/Equality

<sandro> csma: we don't have an unambiguous translation to XML, so we can't consider this yet.

UCR

christian: which plans and changes?

adrian: many of the use cases need more expressive dialects than the ones we have at the moment.
... for example 4.3
... first rule uses negation, which is not in BLD.
... there is many different negations, recall: http://www.w3.org/2005/rules/wg/wiki/negation?highlight=(Negation)

christian: not include WRONG language in the document.

<Harold> I plan to rewrite the relational factorial example (http://www.w3.org/2005/rules/wiki/Equality) to a PS version and complement it with a functional version.

christian: remove "non-encodable" code examples

Leora: 4.5
... modalities not easily expressible.
... also would need negation.

Adrian: let's discuss 4.6

Josb: first a question about abridged syntax.
... in the examples... what is the type of "holdsAt" "ineffective"? is it a URI, a local constant?
... that is not abridged syntax, needs to be fixed!

Adrian: needs understanding of event calculus.

Chris: Is the intention of this rule expressible in BLD? I think yes.

Leora: shall we rediscuss encodings of action languages, sit.calc., event calc. in BLD? (was discussed some time ago in RIFRAF)

chris: abridged syntax.... we had agreement that we discuss only on syntaxes which people are willing to implement.
... for the use cases, it should be the same rationale. abridged syntax only allowed where we have a translator.

jos: 4.6 and other examples need to be reparied to have unambiguous syntax.

<ChrisW> ACTION: adrian to update examples in UCR to presentation syntax [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action08]

<trackbot> Sorry, amibiguous username (more than one match) - adrian

<trackbot> Try using a different identifier, such as family name or username (eg. agiurca, apaschke)

<ChrisW> ACTION: apaschke to update examples in UCR to presentation syntax [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action09]

<trackbot> Created ACTION-585 - Update examples in UCR to presentation syntax [on Adrian Paschke - due 2008-10-03].

christian: What do we do about the syntax?

<ChrisW> ACTION: apaschke to add a comment in UC 4.6 explaining that the example can be translated to BLD using some kind of encoding [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action10]

<trackbot> Created ACTION-586 - Add a comment in UC 4.6 explaining that the example can be translated to BLD using some kind of encoding [on Adrian Paschke - due 2008-10-03].

<ChrisW> ACTION: apaschke to remove examples in 4.3 that we have no dialect for, and add comment to make that clear [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action11]

<trackbot> Created ACTION-587 - Remove examples in 4.3 that we have no dialect for, and add comment to make that clear [on Adrian Paschke - due 2008-10-03].

<AdrianP> I update the class membership test case

<ChrisW> ACTION: leora to remove examples in 4.5 that we have no dialect for, and add comment to make that clear [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action12]

<trackbot> Created ACTION-588 - Remove examples in 4.5 that we have no dialect for, and add comment to make that clear [on Leora Morgenstern - due 2008-10-03].

dave: volunteer to correct 4.8 BLD transcription

<ChrisW> ACTION: dave to rewrite BLD examples from UC 4.8 [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action13]

<trackbot> Created ACTION-589 - Rewrite BLD examples from UC 4.8 [on Dave Reynolds - due 2008-10-03].

Gary: internaitonalization hasn't yet been discussed.

axel: does that just mean we should have a UC/TC which uses rdf:text?

<sandro> Sandro: something that shows off the use of language tags.

<ChrisW> ACTION: apaschke to add a requirement that is satisfied by rdf:text [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action14]

<trackbot> Created ACTION-590 - Add a requirement that is satisfied by rdf:text [on Adrian Paschke - due 2008-10-03].

COFFEEBREAK!

<_mdean> scribe: Mike Dean

<_mdean> scribenick _mdean

<_mdean> nick mdean

<mdean> scribenick: mdean

reviewing actions

NOT CP = NOT CRITICAL PATH

ISSUE-26: roundtripping

<trackbot> ISSUE-26 Replication of original rules after roundtripping to RIF [NOT CP] notes added

Harold: BLD#Conformance_Clauses addresses roundtripping

ChrisW: issue addresses non-semantic things - semantics are required
... metadata SHOULD be preserved
... ready to close issue?

josb: not sure what it means for metadata to survive roundtripping
... e.g. ordering of conjunction clauses

csma: talking about explicit metadata defined in spec

josb: should be made explicit
... talking about conjunctions within metadata

csma: metadata about metadata

ChrisW: who would really care?

csma: WG should not specify such a mechanism
... should forbid subsets of Core
... otherwise notion of Core disappears

ChrisW: profiles different than dialects

Sandro: only WG can say what RIF is

ChrisW: profile provides way to describe what subset of a dialect you support
... issue predates BLD
... what was Core is now BLD

csma: no mechanism for profiles and don't allow subsets of Core

Sandro: constraints on us vs others

Gary: forward- vs. backward-chaining example
... how high is bar for translator implementers

csma: will happen - where WG should legislate is different issue

Harold: cannot predict which other subsets of Core will be relevant (e.g. OWL RL)

DaveReynolds: Core document doesn't currently require all dialects to implement
... suggest no to profiles - informal seems good enough - add later if demand warrants

ChrisW: no objections to this as proposed resolution

<sandro> PROPOSED: Close issue-29 saying we will not define a notion of Profiles

<sandro> PROPOSED: Close issue-29 saying we will not define a notion of Profiles (ie subsets of defined dialects, which are not themselves dialects)

csma: wait until tomorrow for resolutions
... like waiting until the next telecon

<Harold> We MAY introduce Best Practices for RIF. Example: Attach annotations only to syntactic parts of rulesets that stay unchanged under semantics-preserving ruleset transformation. Rationale: You don't want to lose "annotation attachment points" under such transformations (cf. Gary's/Jos' tautology example).

<sandro> PROPOSED: We will not constrain ourselves about whether or not there will ever be a RIF dialect which is a subset of Core

ChrisW: should we constrain the WG?
... proposing inverse of that

Harold: should not constraint ourselves

most people feel we should not

<sandro> No objections, but CSMA wants to think about it more.

ChrisW: these 2 resolutions will close issue 29

<sandro> Chris: passing those two resolutions tomorrow will close issue-29

Issue 33: specification of data sources in RIF

csma: special session later, also on 37 and 38

Issue 39 and 46

ChrisW: 46 subsumes 39
... Michael had more general thing in mind
... can we close 39 because we have it in BLD and then move on to more general

csma: discussed w.r.t. Core? May be a problem for PRD
... semantics of rule set is in conflict resolution strategy - different CRS could be a problem

josb: PRD could preclude loading ruleset with different strategy
... maybe change issue

csma: if issue for PRD, then also an issue for Core

ChrisW: keep open for Core and PRD

Issue 46: modules

csma: related to issue 33

Michael: should be in FLD - can't substantiallly change BLD now

josb: add to FLD and write specialization text for BLD

Michael: should leave this open

ChrisW added note that modules apply to FLD

Issue 50: Semantic metadata

ChrisW: opened for metadata that impacts semantics, e.g. import or PR priority, resolution strategy, negation semantics

Michael: problem for FLD not BLD

ChrisW: closed for BLD

Harold: Peter pointed us to that issue

Michael: metadata is annotations, not import
... FLD has no provision for metadata affecting semantics

DaveReynolds: in logic language, would make it part of language not metadata

<sandro> so the question is what is metadata.... annotations....

Michael: e.g. overrides predicate on identifiers
... depends on where you put stuff

ChrisW: need to decide for FLD, already decided for BLD, what about PRD and Core?

csma: still open

Sandro: would prefer that metadata not be semantic, by definition

csma: point of preserving during round-tripping

ChrisW: how long would this discussion take?
... any objections to precluding semantic metadata?

<sandro> michael: I don't feel comfortable closing this right now, for FLD.

Harold: divide annotations into pragmas and ...

moved to Technical Design product (no conjunctive drop down lists)

Issue 57: Extensibility

ChrisW: XTAN has no official status

csma; relationship to 69

ChrisW: have to close or postpone all issues by Last Call

not in BLD for Last Call

Sandro: can we meet our requirements without it?

<sandro> Sandro: I don't think we can meet our requirements with out.

csma: on tomorrow's agenda?
... proposed resolution for profiles

ChrisW: just roundtripping
... return to issues in last session today

Issues 37 and 38

Christian: references to external classes such as Hen

josb: Jim the Hen Handler

DaveReynolds: schema vs. object model (37 vs. 38) - raised but formal spec not written down
... reuse JAXB mapping

Sandro: write a spec that JAXB happens to implement

Gary: map complex classes to frames
... JAXB is long and complicated - start with something simpler like Java Bean
... handle 1-to-1 vs. set cardinality constraints
... unique properties would probably need equality in the head
... perhaps end up with mini ontology language

DaveReynolds: XML Schema provides cardinalities
... just need to know URIs to reference complex classes
... spec'd algorithm vs. annotation mechanism

csma: such schemas already exist

DaveReynolds: focus on algorithm

Gary: JAXB handles most schemas, but it's a huge spec

csma: useful to any dialects?

Gary: seems orthogonal

csma: who could write strawman document?

Sandro: how many people would use this? is it critical?

Gary: yes

csma: agreed

Adrian: what about black box model?

csma: requires sharing XML Schema + mappin

ChrisW: black box still requires dealing with uniqueness of slots in frame syntax and access to objects

<BobMoore> Just to announce I am here - at least for a little while until I have to go down to supper

<BobMoore> Is it possible to activate the phone bridge

csma: multiple models mapping on to same XML Schema
... navigating schema, attribute of sub-sub-element
... user-defined builtins for object model methods
... how do you refer to methods
... using frame syntax
... e.g. for Java objects

<sandro> Hey, BobMoore - one minute.

DaveReynolds: very PRD specific

<sandro> BobMoore, it looks we're just about to break for lunch.

<sandro> back in 80 minutes or so.

ChrisW: meet-odds

<BobMoore> What timing!!

ChrisW: any other issues or barriers?

csma: use the schema to navigate the data

<sandro> ACTION: Gary to draft a straw proposal addressing part of ISSUE-37, in the area of navigating the schema/data. [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action15]

<trackbot> Created ACTION-591 - Draft a straw proposal addressing part of ISSUE-37, in the area of navigating the schema/data. [on Gary Hallmark - due 2008-10-03].

Gary: not going to do arbitrary XML

csma: use XPath?

Gary: no

<sandro> ACTION: csma to open issue based on the White Board line: "What about methods -- Ignore" [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action16]

<trackbot> Created ACTION-592 - Open issue based on the White Board line: \"What about methods -- Ignore\" [on Christian de Sainte Marie - due 2008-10-03].

break for lunch

cafeteria somewhere in this building

2 or 3 restaurants in Trump Building

appears to have stopped raining

Sony building also has food

reconvene at 2pm

<csma> Bob, we restart in 5 minutes, with about 1 hour on Test cases, then reframing the semantics of conditions in PRD to align it with BLD.

<csma> Ping us if/when you want to join

<BobMoore> okay trust you had a nice lunch - I am assuming the conference code is 74394

<BobMoore> I'm ready to join

<csma> ok

<csma> Call Zakim (+1.617.761.6200 (US), +33 4 89 06 34 99 (F) or +44.117.370.6152

<csma> (GB)

<csma> Conference code is as usual: 74394# ('RIFWG')

<csma> We are setting up the communication on our side

<csma> Bob, the code is 26631, instead

<csma> Scribe: stella Mitchell

<csma> scribenick: StellaMitchell

<sandro> BobMoore, we're trying to figure out how to get the phone working.

<csma> Bob, are you +0777841aaaa?

<BobMoore> yes I am - Zakim says I am the first participant and I need to wait for others to join

Test Cases, including FPWD plan

csma: start with FPWD plan
... what do we need to publish?
... document does not currently include the test cases. Should it?

<sandro> http://www.w3.org/2005/rules/wiki/Test

<AdriaP> http://www.w3.org/2005/rules/wiki/Test

Sandro: there are 3 options:
... 1. owl pulbished the tests, but not the format
... 2.currently for RIF we have the format, and other information, but not the tests themselves
... 3 some groups do neither
... I can't see reasons to go one way or the other

csma: do we need to publish anything? or can we just maintain the wiki page?

jos: need versions

<ChrisW> who is on the phone?

sandro: can't run test cases from wiki, I am writing software to extract the cases from that

adrian: I think it's good to have a separate document also

csma: but does it need to be on rec track?

jos: I don't think it needs to be rec track

csma: needs to be accessible, but rec track is a different quesiton

leora: what do other groups do?

sandro: owl did, but they were different because they included the tests and conformance clauses

csma: if we do not publish the test cases, then this can be on the same web page

sandro: I don't think it needs to be rec track
... people may notice it more if there is a document

csma: question of whether there is a document is separate from the question of whether it is rec track

sandro: I like that the tests are normative, so you know if someone fails a test case then they have a non-conforning implementation

jos: there could be errors in the test cases

harold: there could be errors in the test cases, but it would be good if they are normative - and we should try to make sure they are all correct

csma: if we make them normative, then that doesn't add much work to what we have to do anyway
... but other things, such as running the test cases, are a distraction that is not high priority
... how far are we from being able to publish a document if we put all the tests in an appendix?

adrian: I think it is almost ready

csma: how much of the document has been discussed and agreed upon?

harold: we need to bring it to a high quality anyway - so that they can be used

csma: how hard would it be to agree on properties?

sandro: it's not done yet
... should extend rdf and owl
... I think
... if the definitions are the same as RDF/OWL used

chrisw: is everyone ok with doc being published as WD, and including actual test cases?

dave: nervous about including test cases in the document

sandro: the doc could end up being too large

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

sandro: a recommendation can be multple web pages

adrian: show/hide buttons

sandro: is it ok to use RDF for the manifest file format?

<Harold> In response to Sandro, if the document is too long to be printed, it should carry a label warning about this ("Please don't print this long document ...").

sandro: question to Gary, is this ok for you? will you have tools to parse RDF?

Sandro wants tests to be maintained on the wiki

not in a repository like the current document specifies

Gary: I don't think it's a problem - re: manifest file format

<sandro> sounds like it's okay to be RDF-centric in test-case management for now.

<sandro> SPARQL test cases --- looks like a WD, but it's not.... http://www.w3.org/2001/sw/DataAccess/tests/r2

<sandro> chris: prefer to have text of test cases in WD --- 8

<sandro> chris: prefer not to have it in --- 3

chris: what other things do people expect to see in the WD that is not here
... none were identified
... so, we need to work out properties and include text of test cases

<Harold> Dave, we could have an "In doubt leave it out" policy for the set of approved WD test cases.

chris: test cases task force will meet monday 11:00 est
... have estimate by tues of how long it would take to get this ready to be reviewed as a WD

<ChrisW> ACTION: csma to put Test Cases working draft schedule on agenda for next telecon [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action17]

<trackbot> Created ACTION-593 - Put Test Cases working draft schedule on agenda for next telecon [on Christian de Sainte Marie - due 2008-10-03].

toplic: reviewing test cases

<sandro> subtopic: http://www.w3.org/2005/rules/wiki/Equality_in_conclusion_1

csma: any problems with this

chris: could use more in the description

<sandro> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Equality_in_conclusion_1

<sandro> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Equality_in_conclusion_1

<ChrisW> ACTION: Stella to add more to the description of use case equality in conclusion1 [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action18]

<trackbot> Created ACTION-594 - Add more to the description of use case equality in conclusion1 [on Stella Mitchell - due 2008-10-03].

csma: equality_in_conculsion2
... any objections?

<sandro> Chris: It would be nice to have some real-world example for this.

chris: for description - say why you would write a rule like this
... test cases are examples too

<sandro> Chris: put in a concrete example....

<sandro> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Equality_in_conclusion_2

<sandro> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Equality_in_conclusion_2

<ChrisW> ACTION: Stella to add more to the description of use case equality in conclusion2 [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action19]

<trackbot> Created ACTION-595 - Add more to the description of use case equality in conclusion2 [on Stella Mitchell - due 2008-10-03].

csma: equality_in_condition
... in this test case, the conclusion is exhaustive

jos: no, it's not

sandro: would be nice to see what isn't entailed, as well as what is entailed - n the same test case

various: in this case, the conclusion is not a proper document

dave: updated to be a document

<sandro> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Equality_in_condition

<sandro> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Equality_in_condition

<AxelPolleres> what about testcases like modeling equalities from OWL - e.g. by inverseFunctionalProperties... - for head equality? do we want that? I have one here: http://axel.deri.ie/~axepol/presentations/20080922KeynoteXinnovationsPhDworkshopBerlin.pdf, slide 31, with the test data from slide 22, basically.

<AxelPolleres> stella, let me know, if that looks worthwhile, I can shape it in the right format.

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Guards_and_subtypes

<ChrisW> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Guards_and_subtypes

axel, yes that would be good

<ChrisW> ACTION: josb to update description of Inconsistent entailment [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action20]

<trackbot> Created ACTION-596 - Update description of Inconsistent entailment [on Jos de Bruijn - due 2008-10-03].

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Inconsistent_Entailment

<ChrisW> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Inconsistent_Entailment

test case individual data separation inconsistency

chris: is this still true?

jos: a is a class, and then we say everything is of type a

chris: is this still true in OWL2?

jos: yes, I think it does

chris: in owl1.1 you could use same iri as both a class and an instance

jos: oops, I was talking about a different test case
... in owl-dl there is a separation between individual domains and data value domains

csma: any objections to this test case?

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Individual-Data_Separation_Inconsistency

<ChrisW> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Individual-Data_Separation_Inconsistency

csma: test case "local constant"

jos: in BLD LC WD, this was true , entailment relation doesn't preserve names of local constants

mk: in the current BLD, this is no longer true

mk, jos: this test isn't valid wrt to the current BLD

csma: test case: named_argument_uniterms/frames1

daver: tihs case, together with the other named_argument uniterm/frame give the idea
... can cross link between the 2 tests

jos: for 2nd one, the conclusion is not a valid question
... ...because a predicate can only occur in one context

<sandro> someone should produce a Negative Syntax Test from this: ex:p(ex:a->1 ex:b->2) and ex:p(ex:a->1)

jos: even if it is in a separate document

csma: repeat the arg names

mk, dave: no, not allowed

csma: if rhis cannot be expressed in RIF, why do we need named arguement uniterms?

daver: I kind of agree

mk: can we go back and remove restrictoins such as no polymorphic symbols

chris: would be nice to capture named-argument uniterms as a negative syntax test

dave: will do it now

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Named_Argument_Uniterms/Frames_1

sandro: and change the names of the tests

<sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/No_polymorphic_symbols

csma: test case: no polymorphic symbols
... syntax error - rejected by consumer

sandro: description should highligh that buy has 3 parms in one place, and 4 in another

<ChrisW> ACTION: Stella to update description in http://www.w3.org/2005/rules/wiki/No_polymorphic_symbols [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action21]

<trackbot> Created ACTION-597 - Update description in http://www.w3.org/2005/rules/wiki/No_polymorphic_symbols [on Stella Mitchell - due 2008-10-03].

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/No_polymorphic_symbols

<ChrisW> RESOLVED: accept http://www.w3.org/2005/rules/wiki/No_polymorphic_symbols

<sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/No_polymorphic_symbols

csma: test case: non-annotation entailment

jos: this is the complement of annotation entailment test case
... in the earlier test case we were talking about owl-dl annotation entialment and this is plain owl-dl enatailment, so annotation properties are not considered in the entailment

chris: why do we need to prevent entailment of annotation properties?

jos: just following what is in owl

<ChrisW> ACTION: josb to update description of http://www.w3.org/2005/rules/wiki/Non-Annotation_Entailment [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action22]

<trackbot> Created ACTION-598 - Update description of http://www.w3.org/2005/rules/wiki/Non-Annotation_Entailment [on Jos de Bruijn - due 2008-10-03].

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Non-Annotation_Entailment

<ChrisW> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Non-Annotation_Entailment

csma: test case: owl combination vocabulary separation inconsistency1

jos: again, mixing of individual and concrete domains

axel: this would not be syntactially correct in owl-dl

various: but this is RIF
... combination with OWL-DL

csma: axel, do you object?

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/OWL_Combination_Vocabulary_Separation_Inconsistency_1

axel: no

<ChrisW> RESOLVED: accept http://www.w3.org/2005/rules/wiki/OWL_Combination_Vocabulary_Separation_Inconsistency_1

<AxelPolleres> my concern/ confusion basically was along the following lines:

sandro: will these (the ones applicable to core) turn into core tests?

test case; owl combination vocabulary separation inconsistency2

<AxelPolleres> e.g. the pD* entailment rules from ter Horst 2005 would have no problems with http://www.w3.org/2005/rules/wiki/OWL_Combination_Vocabulary_Separation_Inconsistency_1 if written in RIF.

chris: is this still in owl2, object properties, datatype properties...?

jos: yes
... but in owl2 you have punning, but you can always determine from the context

dave: I thought that was dropped for object properties

<AxelPolleres> ... it is just that I was astonished - hadn't realizzed that - when something is not in OWL DL syntactically we infer inconsistency... but if that is the agreement - and I don't have a better suggestion - then that's fine.

<sandro> recent change in OWL2: "The major change to this document since the version of 11 April 2008 reflects the major revamping of the functional syntax to disallow punning between classes and datatypes and between object, data, and annotation properties. Some minor changes were made to reflect changes in the Functional Syntax."

axel: punning is not trivially combinable with our rules

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/OWL_Combination_Vocabulary_Separation_Inconsistency_2

<ChrisW> RESOLVED: accept http://www.w3.org/2005/rules/wiki/OWL_Combination_Vocabulary_Separation_Inconsistency_2

test case: frame slots are independent

scribe: renamed from named/arg untierms...

csma: any objection to accepting this?

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Frame_slots_are_independent

<ChrisW> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Frame_slots_are_independent

<ChrisW> PROPOSED: accept http://www.w3.org/2005/rules/wiki/Named_Argument_Uniterms_non-polymorphic

<ChrisW> RESOLVED: accept http://www.w3.org/2005/rules/wiki/Named_Argument_Uniterms_non-polymorphic

csma: who read the description of semantics of conditions in PRD?
... a few
... in 1st WD, the semantics was described operationally
... question was rasied of how we relate this to DTB?
... suggestion was to describe semantics of conditions in PRD in the same was as in BLD

chris: I don't understand the link, between need to link to DTB and how to specify conditions
... question about what is being done here, with actions vs. conditions

gary: prd: start with rules and facts, then ask which rules you want to fire
... conditions tell you which to fire (might only pick some of them)
... the rules that fire cause actions to be executed
... actions change state of things
... then repeat

csma: semantics of facts in actions is the same as semantics of facts in conditions
... adrian proposed a definition that is as similar to BLD as possible

gary: it is possible to just point to BLD from here?

csma: Definition of Satisfaction is the important point

gary: why do we need extra substitution step mapping variables to ?

csma: we wanted to mention this in the plenary so that semantisists can comment

jos: why don't use you a semantic structure?

csma: we do
... but we also have operational semantics for the action part

jos: the set of facts is equivalent to a semantic structure, and go directly from w to w'

csma: I'd like to see it written as a draft, and then we have to relate it to the working memory
... and take into consideration the target audience

gary: "w" is very syntactic

,,,2 different things in syntax map to same thing in the domain

adrian: and it needs to be restructured

csma: no, I moved it into the "instantiate rules" part
... ...but I agree that the document needs to be reorganized

jos: you need the substitution because of assert and .. in head

chris: not if you define it as transitions

jos: ok

csma: we can in one of several ways, but currently it is incorrect
... (definition of pattern matching)
... get rid of sigma

.chris: target audience may not understand what the variable mapping means

csma: we need to add a paragraph to explain variable binding
... in signature, adrian introduced a function "type"

adrian: that is to introduce a multi-sorted type system

csma: I'm not sure this is required for PRD

gary: you can do type referencing with guards

csma: and a few other things that could be removed from the signatures

daver: and another mistake: frames don't have an arity

gary: this is supposed to be mirroring BLD in the syntax area

csma: but syntax is what we will define semantics of
... and other simpler things

<ChrisW> taking a break

csma: the main point was about the semantics of pattern matching

<ChrisW> ACTION: AdrianP to reformulate PRD condition semantics without substitution [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action23]

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

<ChrisW> ACTION: AdrianPaschke to reformulate PRD condition semantics without [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action24]

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

<ChrisW> ACTION: AdriaP to reformulate PRD condition semantics without [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action25]

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

<ChrisW> ACTION: APaschke to reformulate PRD condition semantics without [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action26]

<trackbot> Created ACTION-599 - Reformulate PRD condition semantics without [on Adrian Paschke - due 2008-10-03].

<BobMoore> Guys & gals I'm falling asleep - it's been a very long day. I'll try and be a more active participant tomorrow. Enjoy dinner

<sandro> Enjoy, BobMoore

<Blaz> scribenick Blaz

<Blaz> scribenick: Blaz

chrisw: open issues in BLD: roundtripping; at risk: external frames (because there was confusion about exactly what they were), equality in the conclusion, scrictness requirement @ conformance

<StellaMitchell> ScribeNick: Blaz

chrisw: at the last f2f we discussed the idea of 'consumer' that implements extensions; we added the clause there may be a strictness mode that will reject everything that is not in BLD

csma: the point is it is not a requirement

chrisw: there is no harm in leaving "at risk" in for now

csma: equality in the conclusion remains 'at risk', because some of the implementers might want not to implement it

jos: it might be possible to introduce an inconsistency even without this

mk: nobody will implement this anyway
... we should do something about conformance, then

csma: we can not remove it now

<josb> also with RDF and RDFS combinations we can have inconsistencies, because there can be inconsistencies in these languages

what is the terminology - head/body? condition/conclusion?

chrisw: the language should be made consistent

mk: 'premise' is used a lot, 'body' is used in a couple of places
... 'body'/'head' is used in the translation to XML too
... conclusion and premise?

voting on "premise" and "condition"

condition wins

<sandro> ignoring Axel and Sandro saying "if-part"

<sandro> not really a WG decision.

<sandro> Chris: At very least, get rid of Body and Head.

chrisw: leaving equality at risk
... what is external about external frames

mk: it's just like external predicates... it's just the same thing, but different syntax

csma: P(a,b) .. if P has a fixed interpretation, then this is an external: Ext[P(a,b)]

<StellaMitchell> mk: you are asking an external source, and it tells you yes or no

<StellaMitchell> ... you are making a query to an external source

csma: [to mk] is it a way to address 'external methods' ?

Ext[P(a,b)] vs. Ext(a[P->b]) -- is using the same device for differrent purposes

csma: my understanding of the current def. of 'external' is that it is a function

mk: originally, the idea was just to do builtins, then we discussed that people might want to access external datasources

csma: but this is completely different

mk: no

dave: the builtin location is part of a step

mk: you can not really draw a boundary between external sources and builtins

csma: I think it is wrong

mk: it really means a query to some source
... we are just asking a query

csma: I agree they are the same
... if a and P are both IRIs .. which one is external?

mk: if a is an IRI, it means 'query me this way'
... if it's not an IRI, but some other constant .. you would do some other thing .. the object tells you how to query it
... it's a matter of modelling. you can model something as a predicate, relation, class,

csma: in Ext[P(a,b)] there is no ambiguity. what is external is 'P'. if both are equivalent...

mk: I'm saying I can think of situations where you have an object that is sitting externally and that object can be asked various things
... and that object http://www.ibm.com/ and P would be 'president' ..

csma: when I receive a document where there is an external frame and I have to deserialize it .. how do I know which is extenral? I know when I'm executing it, but how do I know when I'm deserializing it ?
... this is an interchange format; when I see an XML document I want do deserialize it into my own lanugage

mk: how do you perform calls to external sources?

csma: I'm not saying we do not need it
... I have to know wether 'a' is stored in an object database at the address given by 'a' or if 'P' is ____. I need to know it syntactically

chrisw: I don't understand the role of externals

mk: if you're getting a document with an ext. and unless you know how to call the external, you can not use it ..

csma: then, what you're saying, it is indeed related to the question about user defined data models and functions. this is what I hear

dave nods no

mk: for the predicate you have to know how many arguments it has. then you also have to know the model of the predicate.

chrisw: the moment you use it, you find out its model. why do you need external?

mk: otherwise you'll have to try to match it to your own predicates

csma: external tells you its interpretation is fixed outside the ruleset
... fixed interpretation is a nice way to say it.

gary: it's fixed but you might not know what it is

csma: it's interpretaiton is external to the ruleset
... to me it is clear if I find an ext. predicate I know there is a specification of that predicate somewhere and I have to know it if I want do do something and it might be an extension if it is a call to a database or an intention of ...
... I can not get what an external frame means

chrisw: the question is: what is external: in Ext[P(a b)] it's P. how about in Ext(a(P->b]) ?

mk: the meaning of this is not a part of your ruleset

jos: could you view this as an external ternary predicate [the second one]
... that has a fixed interpretation

mk: you also know there should be some protocol to say 'send me this stuff..'

<AxelPolleres> "This section introduces the notion of external schemas, which serve as templates for externally defined terms. These schemas determine which externally defined terms are acceptable in a RIF dialect." from FLD, section 2.5

<AxelPolleres> THis means the externally definined schemas are hard-wired with a dialect.

csma: in the second case, even if you have an IRI, where do I send my query?
... I did not say you have to send it there; but if you have that address, then being able to identify P tells you exactly what to do
... I'm not thinking about anything magical

mk: you have an address and there is some XML. if you want to talk to this address, you have to know what message to send. wether there are 2 or 5 arguments. what this message is to be

csma: I perfectly understand this. my question is:
... my understanding was: P unambigously identifies this address

mk: no

csma: how do I know this address?

chrisw interrupts the discussion

chrisw: who understands external frames?
... jos you try to explain it

csma: give me an example

jos: I have no idea what they are good for

axel: I think they are completely redundant
... which external things are allowed in External? it is defined in RIF DTB in coherent set of schemata. it is fixed for a dialect.
... the dialect has to define what the fixed semantics of this is

csma: the point is, your dialect has to specifiy a consistent schema and define what it means
... BLD does not specify any consistent schema for external frames
... if you want to have them, you have to provide a schema

chrisw: the claim that 4 people just agreed to is that external frames are useless
... this only has use if you're extending BLD with your datatypes

csma: I did not think of that you have to provide a consistent schema

chrisw: so maybe extending a definition of external frames .. just a little note: "meant for extending datatypes for use with BLD"
... don't ever question understanding ...
... does it make sense to pull the 'at risk' statement?
... would you object to removing external frames?

mk: you have one KB somewhere and this tells you how to represent things

axel: if that is a datasource that allows you to query frames, then you'd need to give the address anyway; so you couldn't pack it in an external schema

mk: we are going to a different level
... we can express everything with predicates and do not need frames at all ...

axel: but frames are convenient

mk: the question is 'do we need frames at all or not' -- it's the same discussion

chrisw: anyway. let's stop.
... so we got to the point of 'there would be objections to removing external frames'
... anyone against removing 'at risk' ?

noone against

chrisw: let's add this to the list of resolutions to be passed tomorow

: remove at-risk for external.

chrisw: roundtriping

mk: did we decide on at-risk on equality?

chrisw: we're keeping it

<AxelPolleres> Why then not allow any "parametrizable string template" for external schemata? by that, we could e.g. for free get external calls to SPARQL endpoints or SQL databases, etc. etc.

csma: did we decide to keep at risk for 'strictness conformance'?

<AxelPolleres> ... That would generalize the external schemas in a very beneficial way!

chrisw: no

<AxelPolleres> ... e.g. External( ?X ?Y "SELECT ?X FROM WHERE { ... ?Y ... }" ) would be a possible schema.

csma: we kept it because we'll wait for implementations

<AxelPolleres> ... whereas we currently enforce "RIF term syntax" for the external calls, which is unnecessarily restrictive.

chrisw: the only place we talked about roundtripping is in conformance section
... jos complained that 'what does it mean to maintain the semantic meaning' is not specific enough

jos: this whole paragraph can be removed; it doesn't say much

chrisw: it's trying to say that conformance does not mean things need to look the same

jos: this paragraph does not say anything about conformance

csma: [reads] "a nonconformant implementation might not preserve the semantics in the roundtrip"

jos: does anyone want to have some paragraph about roundtripping?

csma: the question was should we say something about survivability of metadata? this is the important point

chrisw: we could add metadata survivability here [a couple paragraphs earlier]

csma: in some test cases there is a notion of 'the same ruleset' if you send me the ruleset back

jos: the only notion we have now is 'semantics preserving mapping'

csma: if we have that notion of keeping the identity of a document, they can keep the metadata as XML and send it back when they return the document
... wedo not have the notion of the identity of document here
... this is probably why it does not make sense
... our processing model is just producer/consumer. it does not define roundtripping of documents
... so the notion of surviving metadata is not defined as well

mk: it could be a recommendation

csma: we have test cases where the notion of passing metadata along makes sense .. but perhaps not in BLD .. maybe in PRD

harold: it could be just conformant, or annotation conformant
... it is a stronger level of 'conformant'

csma: you have metadata about something that is identified uniquely -- if you use the same identifier for something, you can use the same metadata

jos: it is attached to structural elements

chrisw: here is the proposal: I proposed we remove this paragraph because it does not say anything
... and add here the intention of 'metadata survivability' to [first bulletpoint in RIF-BLD specific clauses]
... "conformant producers and consumers ... should ... preserve annotations ... where possible"
... anyone opposed?

noone.

chrisw: for tomorow, we will vote on this resolution

<Harold> Annotations ''should'' survive BLD round-tripping whenever possible.

<Harold> Annotations ''should'' survive BLD transformations whenever possible.

<DaveReynolds> ACTION: chrisw to draft revised metadata conformance wording for BLD [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action27]

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

<DaveReynolds> ACTION: chris to draft revised metadata conformance wording for BLD [recorded in http://www.w3.org/2008/09/26-rif-minutes.html#action28]

<trackbot> Created ACTION-600 - Draft revised metadata conformance wording for BLD [on Christopher Welty - due 2008-10-03].

<AdriaP> logout

- DRAFT -

SV_MEETING_TITLE

26 Sep 2008

Attendees

Contents

Test Cases

UCR

Test Cases, including FPWD plan

Summary of Action Items

Scribe.perl diagnostic output