Admin

<csma> Scribe: Mike Dean

<mdean> scribenick: mdean

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2010Jan/att-0023/rif-mins-19-Jan-2010.html

<csma> PROPOSED: approve minutes from Jan 19

<csma> RESOLVED: approve minutes from Jan 19

no amendments to agenda

Liaisons

Sandro: SPARQL has a new round of working drafts

ChrisW: status of contacting OWL 2 RL people of implementing SWC

<ChrisW> ACTION: chris to contact OWL2-RL folks about SWC [recorded in http://www.w3.org/2010/02/02-rif-minutes.html#action01]

<trackbot> Created ACTION-973 - Contact OWL2-RL folks about SWC [on Christopher Welty - due 2010-02-09].

Sandro: SPARQL might be last round before Last Call
... started from a different place than RIF

Actions review

ChrisW recording action updates

<ChrisW> close action-972

<trackbot> ACTION-972 Update core, bld, and fld xml schema to reflect resolution on imports closed

close action-879

<trackbot> ACTION-879 Contact SRI. closed

close action-880

<trackbot> ACTION-880 Contact David Jones (Boeing). closed

Public comments

no new public comments

no news on public comments

xml:base in RIF/XML

csma: public comment from Thomas Krekeler

<csma> > = xml:base =

<csma> > The xml:base specification ([4]) does not define on which URIs contained in

<csma> > an XML document the resolving mechanism against their base URIs applies.

<csma> > What about

<csma> > * the 'type' attribute of 'Const',

<csma> > * the content of the 'location' element,

<csma> > * the content of the 'Const' element if its type is xsd:anyURI or rif:IRI?

<Harold> Please have a look at our 28 Nov 2009 email: http://lists.w3.org/Archives/Public/public-rif-wg/2009Nov/0065.html

Harold: independently identified same problem in November
... Const only, per email, never got a reply
... up to spec, RIF handling of xml:base modeled after that in RDF spec
... Const is no longer used in location

Sandro: location is probably most important place to use xml:base

csma: makes perfect sense for location, unsure about Const

Harold: earlier also used Const for active iris

Sandro: in linked data, all IRIs point to real documents

Harold: we do allow relative IRIs
... presentation base uses xml:base - only place we use an attribute

Sandro: BLD spec only mentions "relative" (twice) w.r.t. xml:base

<sandro> sandro: I read the BLD spec to not saying anything about relative IRIs in the XML, and suggest that in PS->XML you use Base to expand it.

<sandro> csma: lexical space of rif:iri is ABSOLUTE IRIs. So there is no point to xml:base.

<Harold> ""The Base directive provides yet another shortcut: it applies to all relative IRIs, such as "Mary"^^rif:iri and <John>. The Base directive expands these relative IRIs to "http://example.com/people#Mary"^^rif:iri and "http://example.com/people#John"^^rif:iri, respectively

<Harold> (http://www.w3.org/2005/rules/wiki/BLD#Overview)

<sandro> DaveReynolds: the mapping from the XML syntax to the lexical space is where you do absolutizing.

DaveReynolds: no mention of xml:base is probably a bug

<Harold> "While the Import directive is handled by the presentation-to-XML syntax mapping, the Prefix and Base directives are not. Instead, these directives should be handled by expanding the associated shortcuts (compact URIs)."

<Harold> (http://www.w3.org/2005/rules/wiki/BLD#Mapping_of_the_Rule_Language)

mapping from presentation to xml:base in text, but not in table

<sandro> PROPOSED: We clarify that relative IRIs are allowed in RIF syntaxes (in Const rif:iri, datatypes, location, and profile), and that xml:base is used in making them absolute; the absolute form is seen and used internally, so that's the lexical space.

<sandro> PROPOSED: We clarify that relative IRIs are allowed in RIF syntaxes (anywhere IRIs are allowed, including Const rif:iri, symbol spaces, location, and profile), and that xml:base is used in making them absolute; the absolute form is seen and used internally, so that's the lexical space.

csma: also question about xml:lang - will send note to Harold and Michael

<ChrisW> +1

<LeoraMorgenstern> +1`

<DaveReynolds> +1

<mdean> +1

<sandro> +1

<AdrianP> +1

<Gary> +0

<Harold> 0 (it's already quite clear)

<sandro> RESOLVED: We clarify that relative IRIs are allowed in RIF syntaxes (anywhere IRIs are allowed, including Const rif:iri, symbol spaces, location, and profile), and that xml:base is used in making them absolute; the absolute form is seen and used internally, so that's the lexical space.

csma: not clear in PRD, at least

ACTION csma clarify use of xml:base in PRD

<trackbot> Created ACTION-974 - Clarify in PRD [on Christian de Sainte Marie - due 2010-02-09].

action Sandro ask Axel to clarify xml:base in DTB

<trackbot> Created ACTION-975 - Ask Axel to clarify xml:base in DTB [on Sandro Hawke - due 2010-02-09].

Changes in PRD

csma: PRD had 2 meetings since last RIF telecon
... agreed that refraction and conflict resolution in PRD had bugs
... didn't reflect how various rule systems work
... 1 big change in definition of refraction and conflict resolution
... condition of rules are matched only after each execution block - should be after each atomic action
... consequence: modification can no longer be atomic
... expose intermediate state in PRD semantics
... clarify that modify is a compound action
... 3rd point: at least 1 case not accounted in rule instance - disjunctive condition where disjuncts differ by the value of some constant
... can't distinguish just by variable bindings
... not clear how to handle editorially
... normalization step may not be sufficient

Gary: read latest email - now believe it is sufficient

csma: least change to PRD spec
... believe this is a correction - would like to not require another Last Call
... sure that it will work because it's less abstract and closer to existing engines and RETE algorithm
... less likely to cover non-RETE engine
... would like to use new action RetractAll
... a real addition - not sure whether this would require Last Call

Gary: modify is RetractAll followed by Assert

csma: will talk to Mark Proctor later today
... still following thread
... do we need resolution or something else?

ChrisW: don't see how to avoid another Last Call
... would require changing implementations
... could use shortest allowable time and try to catch up

Sandro: 3 weeks is the minimum

csma: new version should be finished by end of this week

Sandro: need WG resolution to support publication

csma: have a version that's almost ready
... new definition of rule instance is quite complex
... Gary has much simpler version that should be used instead
... would only be an editorial change later

ChrisW: have a short telecon next Tuesday to pass resolution
... 15 minutes just to pass resolution
... can then publish Tuesday or Thursday
... 3 weeks from that date - early March

Sandro: don't need CR period - can skip given implementations

csma: need to complete implementations

ChrisW: try to get all specs in sync

Gary: remove modified no loop test case

Sandro: definitely need test cases

csma: can work on test cases in parallel

ChrisW: pass all resolutions next week
... last call and updated test cases

csma: add RetractAll as well

action csma send email today regarding telecon next week

<trackbot> Created ACTION-976 - Send email today regarding telecon next week [on Christian de Sainte Marie - due 2010-02-09].

Implementations

csma: where are we, what are we missing?
... are we happy with current BLD implementations?

Sandro: partial test report from Gary

will provide test report for SILK - mixed results

scribe: SILK translated to FLORA 2

csma: no test results from Ontoprise
... no news from Jos deRoo
... STI Innsbruck not planning to do any more - no test report

Sandro: need implementation reports, but test results not required

csma: but more credible with (even partial) test results
... PRD has IBM/ILOG and Oracle
... will ask Mark Proctor about JBoss
... Core covered by BLD and PRD
... FLD has Core Answer Set Programming Dialect and SILK Dialect
... partial implementations of DTB
... in IBM
... Gary implements about 3/4 of DTB

Gary: new builtins will be added to product and used by translator
... translator will be sample or open source - not yet sure of release details

<MichaelKifer> I have to leave for a meeting.

csma: CTIC will do open source DTB Java library
... ready in 1 or 2 weeks
... see rif-dev email
... main problem is still SWC
... need to get implementation

Sandro: specifically talking about OWL portion of SWC
... Eye probably counts for RDF

ChrisW: Cleveland Clinic also counts for RDF

DaveReynolds: would need more than OWL RL

csma: exit criteria only requires one profile
... FuXi implementation report says it includes OWL 2 RL

<csma> http://lists.w3.org/Archives/Public/public-rif-comments/2009Oct/0003.html

Sandro: test results would make this more clear

DaveReynolds: implementation report for OWL 2 RL
... not the same as implementating OWL profile in SWC

Sandro: should check owl:imports tests

csma: sandro has existing action

<sandro> http://www.w3.org/2007/OWL/wiki/Implementations

Sandro: consider Pellet or HermIT
... already explored Jena and Oracle

<DaveReynolds> What about SHER?

Sandro: from IBM

ChrisW: SHER team has evaporated

Sandro: seems to be hybrid reasoning - nobody knows how to implement

ChrisW: not that hard - people just don't care
... DL and rule combinations

Sandro: lots of people have implemented SWRL

mdean: SweetRules, Pellet, Protege (using Jess), ...

Sandro: Jess is not better than OWL RL
... Pellet is most likely option - will contact Kendall Clark

DaveReynolds: issue is that SWC OWL-RIF mapping is different from RDF-RIF mapping
... requires different translator
... coudn't work on until April or May

csma: don't have OWL RL profile in SWC, just OWL DL and OWL Full

Sandro: seemed to make sense at the time - don't recall reasoning
... should point to Wiki not CR in sidebar

ChrisW: more external than internal (Wiki) users

<csma> http://www.w3.org/2005/rules/wiki/SWC#Profiles_of_Imports

Sandro: looking for OWL Direct

ACTION Sandro talk to Kendall Clark about possible Pellet implementation of SWC

<trackbot> Created ACTION-977 - Talk to Kendall about SWC implementation [on Sandro Hawke - due 2010-02-09].

csma: lots of pending actions
... urgent if we don't want to leave SWC behind

Test cases

csma: Assert and AssertRetract test cases

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

csma: previously unsure about JRules implementation
... but no more objections
... Retract(?X) retracts all known assertions about ?X
... RetractAll is all values for one slot

Sandro: perhaps RetractValues rather than RetractAll

csma: if don't delete, then can reassert

Sandro: odd that there's no punctuation between Retract and Assert - never have atoms next to each other in BLD
... maybe OK for Action sequence

csma: all use parentheses
... can only have reserved names

Gary: no worse than AND or OR
... no commas anywhere in presentation syntax

<sandro> I cannot keep straight all the different syntaxes and their variations.....
... (obviously)

csma: closed list of actions with pre-defined syntaxes

<csma> PROPOSED: approve http://www.w3.org/2005/rules/wiki/AssertRetract

<DaveReynolds> +1

<Gary> +1

<mdean> +1

<AdrianP> +1

<sandro> +1

<csma> RESOLVED: approve http://www.w3.org/2005/rules/wiki/AssertRetract

AOB (next meeting)

short meeting next week

next full meeting February 16