RIF Telecon 30 Jan 07

30 Jan 2007


See also: IRC log


csma, Leora_Morgenstern, ChrisW, Deborah_Nichols, josb, Harold, PaulaP, Dave_Reynolds, Allen_Ginsberg, Jeff_Pan, AlexKozlenkov, Axel_Polleres, DavidHirtle, Sandro, johnhall, PaulVincent, igor, Gary_Hallmark
FrançoisBry, MichaelKifer, MichaelSintek, MarkusKötzsch
Chris Welty
John Hall





<ChrisW> Scribe: John Hall

<ChrisW> scribenick: johnhall

<ChrisW> http://lists.w3.org/Archives/Public/public-rif-wg/2007Jan/att-0118/16.01.07-rif-minutes.html

Jan 16 minutes approved

<ChrisW> http://lists.w3.org/Archives/Public/public-rif-wg/2007Jan/att-0112/23-01-07-rif-minutes.html

Jan 23 minutes approved

No agenda amendments


<csma> action-210 closed

Action 210 completed - Allen to add hotels to F2F page

Allen: will there be time on day 3 to show some demos?

csma: time on day 3 for demos needs to be on F2F agenda

ChrisW: people wanting to do demos - contact Allen for infrastructure needed for demos

Allen: telecon facilities - use W3C bridge
... networking - wireless access
... wil look into details

Allen: for accommmodation, recommend Comfort Inn

ChrisW: make reservation now - can cancel later if necessary

<csma> action to allen to check telecon and network for F2F

<ChrisW> ACTION: allen to check telecon and network for F2F [recorded in http://www.w3.org/2007/01/30-rif-minutes.html#action01]

<rifbot> Created ACTION-214 - Check telecon and network for F2F [on Allen Ginsberg - due 2007-02-06].

csma: how long needed for demos?

Allen: maybe 2 hours - depends on who wants to demo

csma: need to know who, for agenda

Allen: deadline Feb 10?

<Sandro> 2 weeks in advance

csma: agenda needed - at least draft - end of this week

Feb 10 OK deadline for demos


johnhall: SBVR - no change

<PaulVincent> PRR - no change

ChrisW: What OMG phase is PRR in?

PaulVincent: V1 to be completed in next few OMG meeting cycles

ChrisW: Is it in finalization?

PaulVincent: to be submitted in April, finalization a further 6 months

Technical Design

<csma> action-182 closed

ChrisW: Discussion - lots of email activity on Sandro's proposal

<ChrisW> regarding the XML syntax from the abstract syntax

Sandro: decouple XML syntax from information that must be known about rules to make interchange possible.
... XML syntax mechanically derived
... 2 classes - fully striped or stripe skipped
... recommend stripe skipped

ChrisW: stripe skipping - discussion now or abstract approach in general?
... stripe skipping seems like implementation

Sandro: people will judge the report from skimming the XML

<Harold> Mapping between fully striped and stripe-skipped is itself an important (though syntactic) interchange transformation.

<Harold> We can 'reach' different classes of languages in this way.

Sandro: the sooner we settle this, the sooner we can implement interoperable tools

ChrisW: we have seen an example a month ago, using abstract syntax
... a lot of discussion and confusion
... now that they have developed - people have seen how abstract syntax can yield an XML syntax - any discussion?
... not on Sandro's specific proposal

Axel: there is a close relationship with OWL ontology

Sandro: I agree, a subset of OWL

Axel: try to convert , have representative classes

<Harold> The fully striped class is important to reach the UML, RDF, ... communities; the stripe-skipped class is important to reach the logic, XML, ... communities: RIF's internal stripe-skipping mapping can help to bring theses classes together.

ChrisW: asn06 or OWL - we need to discuss

<AxelPolleres> Not to be misunderstood: I do NOT suggest or encourage by any means OWL as a syntax for RIF.

csma: we decided to have abstract syntax, but left open how it would be expressed
... for now keep to this, not decide what is normative

ChrisW: how many should be maintained?
... Just one and generate others from it?

Sandro: I did not address this

<Harold> Chris, these two versions could be an example for 'standard dialects'.

csma: it would be trivial if not for stripe skipping?

<AxelPolleres> ... if there is an overlap though, I would welcome it.

Sandro: if not trivial,then straightforward

ChrisW: it seems to make most sense to maintain one

<Harold> The inverse mapping is 'stripe-reconstruction'.

ChrisW: would like to make decision now?

<AxelPolleres> "straightforward" I always only believe/understand after having it written down somewhere. :-)

csma: what should be normative is the abstract syntax
... and how the XML syntax is derived from the abstract syntax

Sandro: Chris, are you assuming a specific mechanism for abstract syntax?

<ChrisW> scribenick: axelpolleres

harold: we should have both asn metasyntax and a fully striped syntax.
... this has advantage of being back and forth translatable between object-oriented and XML world.

<DaveReynolds> translators are not so easy if you want forward compatibility via self-describing syntax

ChrisW: Harold do you want both normative? stripe-skipped and full asn?

Harold: would prefer to call asn "object oriented abstract syntax"

Sandro: straw poll between "meta-model", "ontology", and "abstract syntax"?

<scribe> scribenick: johnhall

Harold: what Sandro did was make a model

csma: a model of RIF
... which is a metamodel of RIF rules
... Sandro's model translates easily

ChrisW:what does that mean?

csma:it is a model of the RIF language and thus a meta-model of a RIF document

Harold: we need a format for interchange

ChrisW:OK to call this approach a metamodel

Sandro: we need to discuss, compared with 'ontology'

Axel: there are metamodel languages - if an ontology, then OWL

<csma> advantage of asn06 over UML is simplicity, as I see it

ChrisW:we are discussing metamodel vs ontology?

Sandro: I was trying to bridge between the two camps
... not argue about UML and OWL

<sandro> bridging between metamodel & ontology via asn06 --- which should map to a subset of OWL and a subset of UML.

<AxelPolleres> fair enough, if it is well-defined in a document as an abstract syntax proposal for OWL.

<csma> neither

ChrisW:I haven't heard any objection to the proposed approach - maintaining one artefact and generating
... are there any objections?

LeoraMorgensternit is not an ontology, just a syntax

<Harold> Leora, is it not an 'ontology' of what rules are allowed, how they look like, etc.?

<LeoraMorgenstern> Harold, no, I don't see that.

<igor> abstract syntax seems fine to me

<AlexKozlenkov> any valid metamodel is an ontology

Axel: not clear on how it is translatable betwenn UML and OWL

<LeoraMorgenstern> One can construct a syntax that defines a language,

<LeoraMorgenstern> and then say, a string does or does not belong to the language.

ChrisW:we do not want to separate the approach from the language?

<Harold> OK, it's not the (model-theoretic) *semantics*, so you are right it is (abstract) syntax.

<LeoraMorgenstern> However, I believe that an ontology does much more.

<sandro> To be clear -- my intent is to help us stay in the intersection of Ontologies and Metamodels.

<LeoraMorgenstern> It organizes objects in a particular way.

Axel: not sure it's worth the effort

<LeoraMorgenstern> I don't see this abstract syntax doing that, unless I have missed something.

ChrisW:asn06 you would not agree, but OWL is OK?

<LeoraMorgenstern> well, sandro, I don't understand what you're saying either --- about the intersection between Ontologies and Metamodels.

Axel:there is no document for asn06 yet?

Sandro: not yet, but I can use the parts of OWL I need

Axel: if you can use part of OWL, RDF, then OK with me
... but we need it written down

<Harold> Leora, right, the RIFRAF ontology is more on the semantic level, but also is making a few syntactic distinctions.

<Harold> Not everything written in OWL is an ontology?!

LeoraMorgensternI don't understand where 'ontology' is coming from

Sandro: to me, an ontology is a set of classes and properties and constraining formulae

<AxelPolleres> +1to "Not everything written in OWL is an ontology", but this is not what worries me.

<Harold> BTW, 'partonomy' could be used when we talk about parts of rules.

LeoraMorgensternthe language is not the ontology

Sandro: the classes are there, and relationships

<Harold> asn06 then is a kind of 'partonomy' language. :-)

<AxelPolleres> my only concern is that RIF is not chartered to do a metamodel language, but well, if we use it, we need to define somewhere, what it means, even if that's trivial.

csma: this is a new topic
... even if we decide to maintain RIF as an abstract syntax/metamodel - maintained & normative
... we also need normative XML syntax

<Harold> I agree with csma: we also need a good XML concrete syntax.

Sandro: agree with csma

Sandro: I am hearing consensus on a two stage approach, but hesitation on stage 1

csma: I hear no objection to maintaining abstract syntax and XML that depends on it
... but we do not agree on the form

<AxelPolleres> I object to maintian an abstract sytax, a concrete synatax AND an ontology, but am not religious on what to drop.

ChrisW: where are you on more formally definiing the notation - OWL and RDFS?

<Harold> csma, the 'form' of asn06 is no problem because it can reach OWL, RDF, etc.

ChrisW: is it reasonable to investigate rather than inventing asn06?

<csma> Axel, if the ontology you are mentioning is the RIFRAF, I think that it is a different story

Sandro: it is too difficult to bridge the gap from OWL without some intermediate form

<csma> Harold, I did no say there is a problem, I said I did not hear consensus on that

<Harold> Right.

Sandro: nailing down all the detail of connecting to OWL is hard work - I do not want to do it unless it is really necessary
... and I would need help. Maybe Axel?

ChrisW: Axel - if abstract syntax is maintained in some arteftact, would you be happy?

<JeffP> +1 if OWL is enough, use OWL rather than asn06

Axel: I tried to deal with abstract syntax to OWL in RIFRAF
... Sandro and I should work together

<DaveReynolds> RIFRAF and rule metamodel seem rather different to me

csma: the big advantage of asn is that it is simple
... if there are convincing arguments that it is adequate, that would be a strong argument for using it

Sandro: we would need to tweak it

csma: even for extensions?

<Harold> Likely, asn06 (in spite of its built-in 'partonomy' features) is a sublanguage of OWL-DL.

Sandro: yes. I have concerns around coactions - no multi-valued properties

csma: is it really difficult to map asn06 to OWL?

Sandro: there is a list restriction wrt OWL Full

ChrisW: most OWL parsers will handle it

<AxelPolleres> does that roundtrip?

Sandro: the only problem with OWL Full is readability of the published RIF

ChrisW: publishing as a UML-like picture would be easiest to understand

Sandro: I'm open to persuasion

ChrisW: if it's just another serialization of OWL
... is this what we are discussing?

ChrisW: if asn06 is just a fragment of OWL Axel would be OK

Sandro: I'm trying to show this

ChrisW: does not matter if it is a fragment of OWL-DL or OWL Full

<Harold> ChrisW, it all started with 'pictures' (F2F breakout presentation informally specifying the CORE in one slide), but we also need a plain-ASCII version, e.g. for copy&paste communication in email bodies.

ChrisW: Axel - is this what you are looking for?

Axel: not my main interest - but would like to know what its official status would be

ChrisW: but you would be OK?

Sandro: is this a resolution?

Axel: if is written down what fragment

<AxelPolleres> ok.

ChrisW: this is the contingency in the resolution - "if asn06 is a fragment of OWL ..."

Harold: we could replace 'metasyntax' with 'metamodel'

ChrisW: concrete syntax will be created from the metamodel

Harold: and partly from the mapping

ChrisW: I agree

<sandro> PROPOSED: we'll maintain the XML syntax(es) of RIF in the 2-step process, where step one will use asn06 (which is understood as being a subset of OWL Full) and step 2 is the mapping from asn06 to XML (striped or stripe skipping or whatever).

csma: if asn06 is fragment of OWL, is the resolution to use it?

<Harold> [Admin] Can a WG have 'extra' results?

ChrisW: yes

csma: would we need to make make the definition of asn06 part of RIF?

Sandro: with asn06 normative and derived content as informative

<AxelPolleres> particularly, any well-grounded KR language in which we can ground asn would be fine with me, not necesarrily OWL ;-) if you write it down into FOL sentences, common logic, KIF, F-Logic, I am also fine

<Harold> 'extra' in the sense that some results can be immediately useful for other WGs.

csma: this is not the same
... the metamodel of RIF and the mapping should be normative

ChrisW: even if generated, the XML syntax is normative

csma: for another dialect, one could generate another XML syntax that is not compatible

ChrisW: any objections to abstract syntax and mapping being normative?

<AxelPolleres> I object, as long the meaning of abstract syntax is not formalized, sorry to be picky

<JeffP> Is the abstract syntax the same as the human readable syntax?

ChrisW: the consequence is one and only one XML syntax

<AlexKozlenkov> it is at least slightly odd

Sandro: derived syntax is normative?
... there is an internal W3C debate on derived intormation being normative

ChrisW: normative/non-normative needs more discussion

<DaveReynolds> Axel - if you have a formally specified mapping from a metamodel to the concrete syntax why do you need additional semantics for for the metamodelling language?

csma: I would agree to Sandro's resolution

<sandro> (repeat) PROPOSED: we'll maintain the XML syntax(es) of RIF in the 2-step process, where step one will use asn06 (which is understood as being a subset of OWL Full) and step 2 is the mapping from asn06 to XML (striped or stripe skipping or whatever).

ChrisW: then we can discuss what would be normative

<JeffP> Dave - but XML has no formal semantics

csma: ... contingent on asn06 being defined as a subset of OWL Full

<DaveReynolds> Jeffp - exactly, we are only using this to indirectly specify a syntax, no addtional semantics is requried

Axel: I want it written down

csma: if asn06 is defined as a subset of OWL Full, we do not have to include it as part of the RIF specification

<Harold> A small point regarding step 2: is it just a mathematical mapping or a mapping that itself is specified in a (W3C-standardized) language?

<sandro> PROPOSED-2: we'll maintain the XML syntax(es) of RIF in the 2-step process, where step one will use asn06 (contingent on asn06 being defined as a subset of OWL Full or some other standard formalism) and step 2 is the mapping from asn06 to XML (striped or stripe skipping or whatever).

<JeffP> DaveReynolds - the need for abstract syntax is usually for defining the semantics, like in OWL

Sandro: ... as a subset of OWL Full or some other acceptable specification?

Axel: OK

<DaveReynolds> Jeffp - we are not talking about asn06 as being about specifying the semantics of rulesets!

Harold: is step 2 just a mathematical mapping, or must it be specified in the normative part?

<Sandro> PROPOSED-3: we'll maintain the XML syntax(es) of RIF in a 2-step ? process, where step one will use asn06 (contingent on asn06 being defined as a subset of OWL Full or some other standard formalism) and step 2 is the precisely specified mapping from asn06 to XML (striped or stripe skipping or whatever).

<csma> +1

<Harold> +1

<AxelPolleres> +1

No objections to proposed resolution

<Sandro> RESOLVED: we'll maintain the XML syntax(es) of RIF in a 2-step process, where step one will use asn06 (contingent on asn06 being defined as a subset of OWL Full or some other standard formalism) and step 2 is the precisely specified mapping from asn06 to XML (striped or stripe skipping or whatever).

ChrisW: OWL and RDF Compatibility - table until later meeting


ChrisW: last week's status - not major for F2F - lots of actions outstanding
... now we have agreement on abstract syntax, relevant to RIFRAF

Axel: I would appreciate comments on proposal sent out

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2007Jan/0105.html

<csma> and thread

<csma> [5] http://lists.w3.org/Archives/Public/public-rif-wg/2007Jan/0106.html


ChrisW: no actions open
... close to next working draft

Allen: I have added csma's section on processes
... now need to chack all the references

Sandro: I have sent pointer to web page for program

<csma> http://burns.w3.org/cgi-bin/wiki_tr

Allen: if the program does the job, there is nothing else to be done

ChrisW: We want a frozen document for the F2F

Sandro: HTML pages are not saved, Allen will need to send them to the group

Allen: I just did it (ran the program)
... I need to link the web page to the Wiki
... will do it in next couple of days

... other co-editors should read it

ChrisW: We are close to the final draft


<AxelPolleres> Just in the context of RIFRAF still: I would like to close or stall action 177, see http://lists.w3.org/Archives/Public/public-rif-wg/2007Jan/0106 mentioned before.

<csma> action-177 completed

ChrisW: action 177 is completed

Axel: but the issue is not solved

ChrisW: the action was to propose

Summary of Action Items

[NEW] ACTION: allen to check telecon and network for F2F [recorded in http://www.w3.org/2007/01/30-rif-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2007/01/30 17:32:11 $

[End of scribe.perl diagnostic output]