Admin

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2010Jul/att-0006/2010-07-13-rif-minutes.html

<csma> PROPOSED: approve the minutes of July 13 telecon

<csma> RESOLVED: approve the minutes of July 13 telecon

liaisons

Actions review

csma: removing string match from xml-data

<ChrisW> close action-1034

<trackbot> ACTION-1034 Contact Stella about test cases status closed

Feedback on SPARQL ER

sandro: a conversation with sparql makes more sense than a big review, because there are several issues
... ask Jos to send comments to sparql comments list

csma: one issue is we think sparql should not restrict RIF

XML data

csma: incorporating Jos' and Michael's comments
... currently doc is in flux
... reduce size of xpath related material

cmsa: use NCName to refer to many attributes and some elements that have no namespace
... no namespace => can't use rif:iri

sandro: suggest using a dummy namespace
... or a local symbol

<sandro> I cant decide if using local is incredibly evil, or pretty clever.

sandro: dummy namespace might be more straightforward
... but could offend some xml sensibilities

csma: will go with dummy namespace for now

Primer

harold: making progress and have received some feedback

<Harold> http://www.w3.org/2005/rules/wiki/Primer

csma: why if a,b rather than if a then b?

<sandro> http://www.w3.org/2005/rules/wiki/Primer#Conjunctions_and_Implications

leora: came from common logic, but is fine with if a then b

harold: don't want to confuse with PRD

<AdrianP> but shouldn't we use a syntax which is understandable to the RIF community in the Primer, since the Primer is meant to give easy access to RIF?

csma: it does not conflict (means the same thing)

chris: I uploaded several editorial changes to the primer on the wiki

harold: we will change to If a Then b

csma: target a review in 2 weeks

<csma> ACTION: Sandro to review RIF Primer by August 27 [recorded in http://www.w3.org/2010/07/27-rif-minutes.html#action01]

<trackbot> Created ACTION-1035 - T oreview RIF Primer by August 27 [on Sandro Hawke - due 2010-08-03].

<csma> ACTION: Gary to review RIF Primer by August 27 [recorded in http://www.w3.org/2010/07/27-rif-minutes.html#action02]

<trackbot> Created ACTION-1036 - Review RIF Primer by August 27 [on Gary Hallmark - due 2010-08-03].

RIF in RDF

sandro: differs from Dave R
... key issue: is RDF a KR?

dave: RIF is different from RDF. An encoding is closed world

sandro: instead of optional elements, use empty lists

dave: isn't this a corner case?

<sandro> dave: I expect extensions to add namespaces, not so much to fiddle with the core elements.

csma: can a dialect mark a mandatory piece of core as optional?

<sandro> sandro: I've gotten used to this idea that rif syntactic elements don't get changed by extensions, even if how they are used (eg their cardinality).

sandro: my mapping of xml to rdf requires that optional or repeated elements be lists

chris: I favor following existing xml syntax (and not lists)

csma: if you want to process a RIF XML document, you must understand RIF

sandro: wants RDF graph (encoding a RIF document) to be "stable" under various RDF transformations
... including open-world processing

<sandro> sandro: It's important to me that people use RDF in an open-world KR sort of way. I don't want RIF-in-RDF to go the other way....

sandro: sparql 1.1 lets you query lists but order is not preserved

<sandro> dave: It's important to me be able to use SPARQL to search for bits of rules -- 1.1 will provide that, so I can live with this.

<sandro> In SPARQL 1.1 you CAN query for list members, but you can't get the elements back in order.

<sandro> (sort of DESCRIBE on the list/rule)

<sandro> dave: You won't write a parser in SPARQL, but it's good to be able to query for rule structures.

<sandro> dave: I'd like the requirement phrased differently, though.

michael: neutral
... probably easier to do some transformations if lists are used

<sandro> michael: The rule encoded in RDF is no longer a rule, so I no longer find the KR argument compelling. But transformations are probably easier with lists.

<sandro> harold: OWL can express some rules, eg subsumption.

chris: using lists doesn't correspond to our xml syntax in an obvious, neat way

<sandro> chris:What sways me against lists is the RDF and XML syntaxes not aligning as well.

sandro: no parallel to OWL2 xml to rdf mapping

<Harold> The names used in the RDF syntax should be chosen as close to those of the XML syntax as possible.

<sandro> the four optional/repeated properties are: directive, sentence, declare, formula

csma: prefer the mapping preserve the rif xml names, but not use lists

sandro: given enough time, would implement with and without lists and see what works out
... but to choose now, would choose lists

<sandro> PROPOSED: Do RIF-in-RDF with repeated properties (instead of the list encoding)

<sandro> -1

<ChrisW> +1

<DaveReynolds> +1

<Harold> +1

<csma> +1

<lmorgens> 0

<AdrianP> +1

<sandro> PROPOSED: Do RIF-in-RDF with the list encoding, instead of repeated properties.

<sandro> +1

<DaveReynolds> 0

<csma> 0

<ChrisW> 0 (prefer cleaner correspondance to XML syntax)

<Harold> 0

<AdrianP> 0

<lmorgens> 0

<sandro> RESOLVED: Do RIF-in-RDF with the list encoding, instead of repeated properties.

<csma> RESOLVED: Do RIF-in-RDF with the list encoding, instead of repeated properties.