RIF Telecon 25-Mar-08 -- 25 Mar 2008

<ChrisW> Chair: Sandro Hawke

<scribe> Scribe: Dave Reynolds

<scribe> ScribeNick: DaveReynolds

<sandro> Recent scribes - Igor 3-18, Leora 4 mar, Stella 11 mar

Meetings

<sandro> PROPOSED: accept minutes of 18 March telecon

<sandro> RESOLVED: accept minutes of 18 March telecon

No news on F2F10

Liaison

Liason

Jos: no news from OWL task force

Sandro: no discussion in OWL wg on this, there is a f2f next week, could push for time on this in the f2f agenda?

Jos: haven't had feedback yet

Sandro: suggest OWL wg find a couple of reviewers for the SWC document

Jos: there will be some changes resulting from changes in BLD, should wait for frozen version

UCR

Axel: f2f10 process is running

<sandro> re ACTION-443 the page is still in the voting stage. Axel, can you make it read better for where we really are.

Chris: updated wiki page a little but needs to be checked and improved

Axel to scribe next week

back to UCR

Metadata

<sandro> (oops, Zakim)

Gary: summary of review, clear that it has not been updated, no link from use cases to the solution, at best it is irrelevant but worse is potentially misleading

<sandro> Gary: UCR is currently very misleading.

Gary: e.g. for the OWL use cases they show direct link from use case to OWL/RDF need similar linkage for the RIF part of the solution

<sandro> Sandro: sounds like more stuff is needed, but is what's there wrong?

<sandro> Gary: now we have solutions to link to. why don't we use the uniform new syntax?

Gary: right. For example, the rule examples are all different syntax should use new presentation syntax
... doesn't link to the technical content of the specs

Harold: the worked examples are not up to date but in several cases we did have some worked examples but more on XML syntax, should look back at these

<Hassan> +1 with Gary all the way

Gary: use a simplified presentation syntax? "Tutorial presentation syntax"?

Harold: would this be better as a primer or separate document?

Gary: no
... just annotate each example with solution from FLD as jumping off point to help people understand what we have done
... RDF and OWL mapping is clear, XML mapping is less definitive but clear enough we can illustrate how the data maps and so what the rules would look like in RIF
... also gives some examples to help people get into RIF without the need for a full blown primer

<Zakim> sandro, you wanted to ask about using covered rule languages

Sandro: makes sense for use case examples to be in vendor language, separately show how they translate to RIF

<Harold> Worked Examples: http://www.w3.org/2005/rules/wg/wiki/D._Evaluation%3A_Use_Cases

Gary: might be nice but too many different languages and the readers won't be familiar with them
... want to teach people about RIF rather than all the languages and how the translation works

<sandro> ack

Gary: some use cases weak, perhaps through those out, leaves maybe 15-20 rules in total

ChrisW: partially agree with Gary but doesn't want UCR to become a tutorial

Gary: not a full tutorial but the OWL one goes somewhat that way

ChrisW: OWL started out as real requirements doc, then refered back and updated that as issues were classified

<Harold> OWL Web Ontology Language Use Cases and Requirements: http://www.w3.org/TR/webont-req/

ChrisW: but for OWL already had DAML+OIL as uniform starting place
... then turned into explanation of the problems that OWL was designed to solve, and so has some tutorial aspect but that's not the purpose

Gary: agreed, move our UCR to closer to OWL style, advertise what RIF is good for

<sandro> http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/0104

Sandro: need to consult with Adrian how much of this task he can take on

<ChrisW> ACTION: chris to send email about UCR [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action01]

<trackbot-ng> Created ACTION-453 - Send email about UCR [on Christopher Welty - due 2008-04-01].

<Harold> RE action 442: http://www.w3.org/2005/rules/wiki/BLD#EBNF_for_RIF-BLD_Rule_Language and http://www.w3.org/2005/rules/wiki/FLD#EBNF_Grammar_for_the_Presentation_Syntax_of_RIF-FLD

Michael: need to be able to provide metadata which affects the semantics of the rule data
... Jos' response is that those are directives rather than metadata

<sandro> semantic metadata vs non-semantic metadata

Sandro: procedurally, thought the group had decide to leave semantic-metadata for later

Michael: yes but need to prepare ground

ChrisW: agreed on simple part at f2f, decided to put that in next WD and continue talking about directives

Michael: but at the moment directives are too simple
... should be able to attach directives to queries, and individual rules, should be keeping this in mind

Sandro: is there an issue being tracked on this?

ChrisW: there is modules, import/inclusion but not a separate one specifically for the directives machinery

<scribe> ACTION: Chris to open issue on semantic metadata [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action02]

<ChrisW> ACTION: chris to open an issue on semantic metadata [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action03]

<trackbot-ng> Created ACTION-454 - Open an issue on semantic metadata [on Christopher Welty - due 2008-04-01].

<ChrisW> ACTION: Chris to stop giving himself actions... [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action04]

<trackbot-ng> Created ACTION-455 - Stop giving himself actions... [on Christopher Welty - due 2008-04-01].

Harold: incorporated the metadata in the BNF grammars (see above URLs)
... next to put it in XML, depends on current discussions
... but concerned that they duplicate some things, e.g. sequences

<sandro> ACTION: Jos to look over Harold's BNF for metadata [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action05]

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

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

<sandro> ACTION: jedbruij2 to look over Harold's BNF for metadata [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action06]

<trackbot-ng> Sorry, couldn't find user - jedbruij2

<sandro> ACTION: jdebruij2 to look over Harold's BNF for metadata [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action07]

<trackbot-ng> Created ACTION-456 - Look over Harold's BNF for metadata [on Jos de Bruijn - due 2008-04-01].

Striping

<Harold> http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/0108.html

Harold: tried to show how this works in graph form in email (see URL above)
... Sandro's version 3 is sort of a reification of the properties (scribe couldn't follow that)

Gary: isn't RDF layout a sort of defn for striping?

Sandro: also Henry Thompson paper on alternating normal form

Gary: striping is interesting because then can treat rules as RDF data and so rules can process the XML representation of other rules
... which would be a reason for defining striping in an RDF compatible way

Harold: could RDF parsers be extended to cope with user defined slots?

Gary: why would you want to that?

Harold: RDF parse could only handle bare bones, but couldn't understand extra attributes inside class forms (?)
... need to fix the RDF schema in order to give an XML schema to RDF (?)

Sandro: disagree, the name of a slot in RIF syntax should be just a literal
... even though fully striped is analogous to RDF we won't be able to parse RIF as RDF due to other decisions (ordering, typing) which he argued against but lost

Harold: slots can in general be terms not just simple literals
... if cannot use direct slots, then tend for a version of solution 3

<sandro> Harold: (if we can't have user-level slots)

[Discussion between Harold and Sandro on whether this is reification or simply a data model, couldn't follow details.]

<sandro> STRAWPOLL: Go with solution 3 in Sandro/Harold's e-mail http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/0108.html

<Zakim> sandro, you wanted to say this isn't RDF/XML for a couple of reasons

<Hassan> I agree with Jos

Jos: favour making syntax slightly less verbose, but not strong opinion

Harold: proposing a *version* of solution 3
... avoids word pair

Sandro: suspects Christian might prefer solution2

Harold: but that has problems, it is positionalizing as two vectors that you then have to match up

<Harold> Going "fully striped" was a long process.

Jos: favour solution1, stop being fully striped, would not object to others

<sandro> http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/0108.html

<GaryHallmark> I'm not sure being 99% striped will serve us well...

ChrisW: propose pick one for next WD, seek feedback from implementors

<sandro> PROPOSED: Use fully-striped syntax for next WD (option 3 in http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/0108.html)

<Harold> +1

<AxelPolleres> +1

<sandro> (not ruling on name "Slot" vs "Pair" vs whatever.)

<sandro> +1

<Hassan> +1

<GaryHallmark> +1

<IgorMozetic> +1

<mdean> +1

<ChrisW> +1

<sandro> RESOLVED: Use fully-striped syntax for next WD (option 3 in http://lists.w3.org/Archives/Public/public-rif-wg/2008Mar/0108.html)

<ChrisW> ACTION: harold to update xml syntax to reflect "option 3" resolution [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action08]

<trackbot-ng> Created ACTION-457 - Update xml syntax to reflect \"option 3\" resolution [on Harold Boley - due 2008-04-01].

Action Review

ChrisW: plan had been to do WDs for end of these week, what is the current reasonable schedule?

Harold: 1 more week

<sandro> Harold -- freeze on Apr 4th?

<Harold> Yes.

Michael: much of datatype text is moved to DTB but this document will take longer to get ready, it has not progressed much recently
... suggest leaving text in place for next WD and put a note saying it will be moved to DTB in future

Axel: so far it is not done but could still be feasible in the next week

Michael: the problem is not the moving, but the fact that DTB is not ready to be published so the moved pieces would then not be included in the WD set

<Harold> Because of this, Axel and I want to have Michael as a third co-editor of DTB, and Michael is fine.

Axel: what about bulitins not decided yet?

Chairs: leave in and mark them

Axel: given that could get DTB cleaned up by end of next week ready for WG review

ChrisW: leave in BLD, postpone deciding on move until see the DTB doc, mark the sections as planned to move

action-425 complete

Action-427 complete

<sandro> syntax example from OWL is: {{EdNote|~~~~|See [http://www.w3.org/2007/OWL/tracker/issues/4 Issue-4] (syntax reordering) and [http://www.w3.org/2007/OWL/tracker/issues/82 Issue-82] (Metamodel diagrams).}}

Action-430 done differently, separated out, not as appendices

ChrisW: people felt the doc should be focussed on helping people implement BLD so the FLD derivation would be better in an appendix for readability
... should in document and still normative, just want it out of normal flow of reading

<Harold> Chris, could a kind of "Readers' Guide" in the Overview do the trick?

<sandro> from F2F9: RESOLVED: make "specialization of FLD" sections (of BLD) appendices, leaving standalone sections in place, and making both normative

<josb> section 2.0.9 were the sub-dialects

Chairs - stick to the resolution or open a discussion to reverse it, starting with argument in email

Michael: semantics of BLD is ready except for builtins, not expecting any changes to the rest

Harold: have added IRIs in the presentation syntax for action-441

Jos: the metadata specification should be in the specification in section 3.4, and it's not there yet

Harold: maybe it is not needed there

Jos: so which bit is the *specification* of the RIF syntax

Harold: the ebnf is almost 1-1 with XML but the "mathematical english" omits the IRIs and metadata

ChrisW: the condition and rule language parts are still separate, would like one place to see whole of the syntax for BLD

<Hassan> +1 with ChrisW

<josb> +1

ChrisW: one place for EBNF, could colour different parts, could duplicate sections but have one place where it is complete

<Hassan> But make sure to keep redundant parts consistent...

<Harold> I guess we want no metadata at places like Rule implication: If f is an well-formed atomic formula and ? is a RIF-BLD condition then f :- ? is a well-formed formula, provided that f is not externally defined (i.e., does not have the form External(...)).

Harold: sections 3.2 to 3.4 are the mathematical english and don't see need for metadata in there

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

Jos: but have different bits of the specification which differ - the mathematical english, the ebnf but no single specification, for example some of the well-formedness constraints are not represented in the ebnf
... so suggest making the mathematical english consistent with ebnf by including metadata and iris, just for completeness

Michael: would require thought, all the syntax for the metadata would have to be defined in that math eng section
... but what is this metadata, they are not terms, what is metadata value for example?

Jos: see the doc, defines as const etc

Michael: but the derivation from FLD is normative so need to decide how and where to add this stuff to FLD. Just needs care.

<Harold> If we add a sentence like "A RIF-BLD rule can have an optional absolute-IRI and zero or more Metadata...", then we could use a different color or font to show clearly that this has no semantic import.

Michael: this is lower priority than semantics for builtins

<Harold> Corresponds to EBNF: RULE ::= 'Forall' absolute-IRI? Metadata* Var* '(' CLAUSE ')' | CLAUSE

<ChrisW> ACTION: michael to add metadata and iris for rules/rulesets to the "mathematical english" [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action09]

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

<trackbot-ng> Try using a different identifier, such as family name or username (eg. mkifer, msintek, merdmann)

<ChrisW> ACTION: mkifer to add metadata and iris for rules/rulesets to the "mathematical english" [recorded in http://www.w3.org/2008/03/25-rif-minutes.html#action10]

<trackbot-ng> Created ACTION-458 - Add metadata and iris for rules/rulesets to the \"mathematical english\" [on Michael Kifer - due 2008-04-01].

<josb> not with me

jos: just want want consistent specification of what the language is

Michael: but the actual language is the XML right?

Jos: but the XML can be defined by mapping from the presentation syntax

<josb> right

Michael: in the semantics section this metadata will be ignored

<JamesMac> http://en.wikipedia.org/wiki/MathML

<JamesMac> Is this still applicable

action-432 is done, see mailing list

action-441 done

<josb> +1

<Hassan> +1

<sandro> ADJOURN

- DRAFT -

RIF Telecon 25-Mar-08

25 Mar 2008

Attendees

Contents