RIF Telecon 15 May 07 -- 15 May 2007

<ChrisW> topic #rif 15 May RIF Telecon, Agenda: Agenda: http://lists.w3.org/Archives/Public/public-rif-wg/2007May/0035.html

<ChrisW> scribe: John Hall

<ChrisW> scribenick: johnhall

next meeting 22 May

no actions

no agenda amendments

Admin

may 8 minutes approved

F2F

f2f6 - all complete survey

sandro: 6 people for dinner but not meeting

<sandro> (I believe those dinner-but-not-meeting people to be erroneous)

josb: f2f all according to plan
... let me know if wiki needs updating

csma: book hotels asap
... f2f7 in New York, IBM september?
... other offers?
... issue with academics - meeting September -first or second half?

<josb> I have no constraints in Sep

csma: preference from Europeans?

<IgorMozetic> no constraints here

<LeoraMorgenstern> This is only my individual constraint, but the High Holy Days fall in September ...

<LeoraMorgenstern> Starting September 13

paula: Summer school in early Sept, several RIF members involved

<LeoraMorgenstern> Which doesn't make a difference if it's in Europe, since I wouldn't go.

paula: http://i-media.tugraz.at/ event in Austria

<Harold> http://i-media.tugraz.at/

<LeoraMorgenstern> But if it would be at IBM in New York, it would be a pity if I couldn't make it.

csma: preference for second half sept?

<ChrisW> ACTION: Chris to send email requesting offers to host f2f7 [recorded in http://www.w3.org/2007/05/15-rif-minutes.html#action01]

<rifbot> Created ACTION-286 - Send email requesting offers to host f2f7 [on Chris Menzel - due 2007-05-22].

<Harold> http://reasoningweb.org/2007/

<scribe> ACTION: ChrisW to email to sound out [recorded in http://www.w3.org/2007/05/15-rif-minutes.html#action02]

csam: 2 weeks to respond

Liason

LIASONS

SBVR - nothing new

Core - Issue 31 (disjoint names)

<ChrisW> action-285: continued

csma: no actions

sorry, have lost sound

<ChrisW> action-284: complete

<ChrisW> do you want to try dialing in again, John?

<ChrisW> action-283: complete

csma: action 283complete

<ChrisW> action-282: continued

csma: action 282 open
... 280, 281 continued
... 276 done
... Dave R did it, no need to recruit someone

lost sound again

zakim P14is me

<sandro> Chris: We said last week that if anyone wanted reflection in the Core, they'd have to argue for it before today.

zakim ??p14 is me

chrisw: close issue on reflection

<sandro> Chris: We didn't resolve preciseley what "reflection" mean. But since no one is proposing a reflection design, I think we should not have it in core.

<sandro> Chris: reflection could even include metadata.

<sandro> csma: That's a property of a rule language, not of RIF

chrisw: if allow to write rules about rules, would be a kind of reflection

csma: not a property of RIF
... extension may allow

<sandro> Chris: Well, if RIF Core allowed you to say something like "all rules written by Chris are false", that's a kind of reflection.

<LeoraMorgenstern> There is reflection that is okay, and reflection that is pathological.

<Harold> Sandro/Chris, I agree we may have a "reflection" dialect, but not in the core.

<LeoraMorgenstern> Chris, your example, of course ,is the Liar's paradox, and is pathological.

<LeoraMorgenstern> But there is a significant amount of reflection that is not problematic.

<LeoraMorgenstern> Though, I would agree with Christian that we can probably live without it.

<sandro> PROPOSED: RIF Core follows OS ("overlapping sorts") (resolving issue-31)

<LeoraMorgenstern> Considering how much else we're living without, and how potentially problematic reflection can be.

<sandro> csma: we may introduce reflection later

<ChrisW> http://www.w3.org/2005/rules/wg/wiki/Issue-31

sorry , sound is very fuzzy

<sandro> PROPOSED: RIF Core follows OS ("overlapping sorts") as on http://www.w3.org/2005/rules/wg/wiki/Issue-31 (resolving issue-31)

maybe someone else could scribe?

<sandro> josb: What if extensions don't want overlapping sorts? Doesn't that mean "extensions" will have to actually be "restrictions"?

<sandro> Chris: Well, but in talking about translations, since you can translate in any direction, that doesn't manifest.

<Harold> Jos, I think 'extension' in a dialect can often mean adding extra constraints, hence 'restricton' (another example: a "recursion prohibition constraint" in a dialect).

<sandro> sandro: we could ask the editor to highlight this issue, but I don't think that's necessary.

is somebody else scribing?

<ChrisW> scribenick: IgorMozetic

<ChrisW> Scribe: Igor Mozetic

<Harold> Jos, maybe introducing a disjointness constraint in a dialect later is related to using http://www.w3.org/TR/2004/REC-owl-guide-20040210/#DisjointClasses

DaveR: Re Issue31 is wondering if he was supposed to provide some wrtiteup

<sandro> Dave: What clinched the discussion last week was Michael Kifer saying there was a translation from OS to DS, given in HiLOG paper

<sandro> Jos: That paper said there *are* cases where you can't do the translation, and these cases are expressible in RIF Core.

<sandro> Chris: These are the cases where you say a predicate is = to an individual?

<sandro> Jos: Yes, non-safe quantification with equality.

<sandro> Jos: This forces all predicates to be interpretted in the same way and leads to this kind of problems.

<sandro> Dave: I think we covered this in e-mail -- only allowing safe rules

<sandro> Jos: I didn't think we were considering such a restriction.

<sandro> Chris: It sounds like we're not really ready to close this.

Chris: Jos to talk to Michael and decide who will summarize the discussion

<sandro> ACTION: Jos to talk to Michael Kifer and have one of your post the results to the list [recorded in http://www.w3.org/2007/05/15-rif-minutes.html#action03]

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

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

<sandro> ACTION: jdebruij to talk to Michael Kifer and have one of your post the results to the list [recorded in http://www.w3.org/2007/05/15-rif-minutes.html#action04]

<rifbot> Created ACTION-287 - Talk to Michael Kifer and have one of your post the results to the list [on Jos de Bruijn - due 2007-05-22].

Core - Metadata

csma: some objections against using RDF syntax for metadata

<sandro> csma: agreement last week to use rdf for metadata -- objection from some folks based on not wanting to implement rdf/xml -- dave said it could be restricted rdf

<sandro> sandro: I said rdf/xml is not the same as rdf --- the plan is to use RDF -- it's not about the RDF/XML syntax.

<sandro> csma: and Gary said he wanted all the elements to be constants, and not terms.

gary: other RDF issues might be problematic: bnodes...

<DaveReynolds> +q

DaveR: one wants to annotate structured values

<sandro> DaveReynolds: sometimes you want to annotate a structured value -- b-nodes are a nice way to do that, without having to mint a URI.

<ChrisW> pointer to use case?

<sandro> DaveReynolds: you could probably restrict those to tree-structure, in which case you could have an XML syntax with no explicit nodes.

<DaveReynolds> use case: http://www.w3.org/2005/rules/wg/wiki/Arch/Data_Sets

Gary: sounds acceptable to me

<sandro> csma: would a dialect have a specific set of metadata?

sandro: we agreed on open metadata

<sandro> Sandro: I can't imagine anyone would say we should exaustively list all possible metadata properties.

csma: should we have a restricted list of metadata?

<sandro> csma: should have a list of metadata properties that are "must-understand"?

<sandro> csma: like "every rule must have a name"

<PaulVincent> Metadata = properties about rules not necessary for rule processing, but may assist in rule management, documentation, further transformations etc (IMHO)

sandro: distinction between metadata and abstract syntax is unclear

<sandro> Sandro: I can't tell the difference between metadata and other parts of the rules.

<sandro> csma: I think it's non-semantic stuff.

sandro: metadata is very tightly connected with the extansibility

<sandro> Sandro: I don't think it makes sense to try to draw a line about which features are "metadata" and which are not.

sandro: all the parts of the rulesets are named by URIs
... this was a result of the Georgia F2F

... except for local names

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

<csma> http://www.w3.org/2005/rules/wg/wiki/F2F5

<sandro> "RIF will use URIs (IRIs) in the style of RDF and OWL, using them to identify at least: globally named predicates, functions, datatypes, constants (OWL individuals, not literal values), rules, and rulesets."

<sandro> -- http://www.w3.org/2005/rules/wg/wiki/F2F4

sandro: maybe we can use RIF itself for the metadata formalism

csma: collect emails (Gary, DaveR) regarding metadata

<sandro> requirements for metadata --- rdf data model (some subset of it, at a least)

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2007May/att-0015/08-rif-minutes.html

<sandro> requirements for metadata -- extensibility of properties

sandro: we didn't decide on the syntax, just the metadata model (RDF)

csma: sandro to write a proposed resolution regarding metadata and send it via email

<sandro> Sandro: is there a difference between using RDF and using RIF ?

DaveR: difference between RIF syntax and RDF datamodel

<sandro> Dave: you could use the triples bit of RIF.

<sandro> csma: something like: RIF metadata will follow the RDF data model and use a RIF-specific syntax.

csma: use RDF detamodel and RIF syntax

<sandro> PROPOSED: RIF Metadata will follow the RDF data model, but not necessarily the RDF/XML syntax

igor: why restrict to RDF datamodel?

<sandro> igor: Why the restriction to binary triples?

<sandro> igor: why not allow n-ary, etc, as in RIF facts in general.

<sandro> Gary: I might be able to live with b-nodes, for things like modification history, or other structured data.

<johnhall> I'm giving up for today - bye

<sandro> igor: I can live with triples-only, but it seems like it would be more natural to allow full RIF facts.

<sandro> igor: such as for structures (where one might use b-nodes)

<sandro> csma: are there many languages that use metadata beyond triples?

sandro: metadata might want to use terms

<sandro> Sandro: I think there may be things that some people might call metadata, for which we'll want to use terms.

<sandro> Sandro: ... therefor I'm not ready to agree to the rdf-data-model-only for RIF metadata.

csma: do we need metadata to add to lower-level entities (like predicates, functions, ...) ?

Core - XML Syntax

Gary: started his XML schema from Harold's DTD
... nothing in the schema for Xpath and Xquery
... just for builtins

sandro: builtins should show up in abstract RIF syntax
... we should track builtins

<DaveReynolds> At F2F5 we did talk about a list of core builtins as being part of the Arch document

sandro: SWRL is a good starting place for builtins

<ChrisW> pointer to "SWRL list"?

<Harold> SWRL Builtins: http://www.w3.org/Submission/SWRL/#8

<DaveReynolds> +1 to Gary, should start with the XPATH/XQUERY functions and operators, as also used in SPARQL

csma: we should start a wiki page for builtins

<ChrisW> Xpath/Xquery functions: http://www.w3.org/TR/xpath-functions/

<sandro> Sandro: I think arch should have a section saying how you import for X/X F&O, and RIF Core should have a list of which builtins we import into Core. And somewhere we have a list of exceptions.

<Harold> relational-builtin(Res, Arg1, Arg2) corresponds directly to Res = functional-builtin(Arg1, Arg2)

<sandro> ACTION: Christian send out homework about builtins and put it on future agenda. [recorded in http://www.w3.org/2007/05/15-rif-minutes.html#action05]

<rifbot> Created ACTION-288 - Send out homework about builtins and put it on future agenda. [on Christian de Sainte Marie - due 2007-05-22].

sandro: asks people to have a look at his Wiki page http://www.w3.org/2005/rules/wg/wiki/Arch/XML_Syntax

<DaveReynolds> Harold - but the precise semantics of how to do things like type conversions may differ, xpath-functions is pretty specific on all that

<sandro> EVERYONE who has XML experience, please review http://www.w3.org/2005/rules/wg/wiki/Arch/XML_Syntax and add your comments.

<PaulaP> +1

<Harold> Dave, right, lets look into this.

- DRAFT -

RIF Telecon 15 May 07

15 May 2007

Attendees

Contents

Admin

F2F

Liason

Core - Issue 31 (disjoint names)

Core - Metadata

Core - XML Syntax

Summary of Action Items

Scribe.perl diagnostic output