RIF Telecon 7 September 2010

7 September 2010


See also: IRC log


Adrian Paschke, Chris Welty, Christian de Sainte Marie (csma), Dave Reynolds, Doug Lenat, Leora Morgenstern, Mike Dean, Sandro Hawke, Stella Mitchell
HaroldBoley, Michael Kifer
Christian de Sainte Marie (csma)
Stella Mitchell



<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2010Sep/0001.html

Christian: Will propose to accept the above minutes at the next call.
... added UCR publication plan to today's agenda per Adrian's request


Christian, Sandro: RIF/SPARQL topics will continue by email

Action Review

action-1047 completed

<ChrisW> trackbot, associate this channel with #rif

<trackbot> Associating this channel with #rif...

<ChrisW> close action-1047

<trackbot> ACTION-1047 Write down "RIF-direct semantics" proposal for RIF-in-RDF. closed

Sandro: I decided not to work on action-1046, so we can close that - it's not high enough priority for the time left

close action-1046

close action-1044

actions 1043 and 1041 are completed

action-1039 closed

action-1040 continued

action-1037 continued

close action-1033

close action-1032

close action-1031

action-1010 continued

action-1004 continued

action-935 continued

action-152 continued

actions 1035 and 1036 continued


Sandro: Axel and I talked about imports and he updated that part of the spec - this will be a Note. It would be good if someone could check over the semantics.
... "rif-used-with-profile" vs. "rif-imported-with-profile" - is Chris OK with this naming?

ChrisW: Yes, I was more concerned that it is past tense rather than "used" vs. "imported".

Sandro: The document needs a review; aiming to publish as a note by the end of the month

Christian: Should we have a telecon each week until the end of the group?

<AdrianP> yes, will make sense to have a RIF telecon for the remaining three weeks

Leora: I will review the RIF/RDF document for clarity and usability since I need to review it in detail in order to write my primer section anyway
... it would take me a week - if I get it by Weds afternoon

Sandro: Remaining changes are not something Leora really needs to wait on

Leora: I'll review it by tomorrow then

<csma> ACTION: Leora to review RIF in RDF by Sept 14 [recorded in http://www.w3.org/2010/09/07-rif-minutes.html#action01]

<trackbot> Created ACTION-1048 - Review RIF in RDF by Sept 14 [on Leora Morgenstern - due 2010-09-14].

Sandro: I would like Dave Reynolds to review this docment also

DaveR: Ok

<csma> ACTION: Dave to review RIF in RDF by Sept 14 [recorded in http://www.w3.org/2010/09/07-rif-minutes.html#action02]

<trackbot> Created ACTION-1049 - Review RIF in RDF by Sept 14 [on Dave Reynolds - due 2010-09-14].

Sandro: Does anyone know about WSML light? - it involves RIF being embedded in RDF as one big string.

<csma> http://www.w3.org/2005/rules/wg/track/issues/105

Christian: We still have an issue open on RIF and RDF: 105 above

Sandro: This has been resolved: that's the recent work by Axel and I, which now needs to be reviewed
... we'll leave it open for now and resolve after the reviews

RIF and XML data

Christian: Almost ready to have the next version of the "RIF and XML data" document reviewed...I think I have addressed all issues raised on earlier versions

<ChrisW> sandro: never believe me when I say "I don't need an action"

<csma> http://www.w3.org/2005/rules/wg/track/issues/103

Christian: Issue 103: spec doesn't refer to that data model any longer

<csma> http://www.w3.org/2005/rules/wg/track/issues/104

Christian:issue 104: using NCNames in spec - issue is really how to deal with attributes and properties that are not in a namespace

Christian: There are 2 suggested approaches:
... 1. use dummy namespace
... 2. extend space of RIF:IRI to contain ncnames
... (1 abuses notion of namespace and 2 abuses notion of iri)

Sandro: Maybe there's a 3rd option...

<csma> o[namespace#attribute(localname)->value]

<csma> o[namespace#localname->value]

Christian: The examples above are what we're talking about: 1st involves attribute of element, 2nd involves child of element

Sandro: Should we spend more telecon time going over this issue now?

Doug: Yes, I think we should

<sandro> ?x[<http://example.org/customertable#Name -> ?y]

<sandro> And( ?x[<http://www.w3.org/XML/1998/namespace#attribute(lang)> -> "en"^^xs:language])

<sandro> ?x[<http://example.org/customertable#Name> -> ?y]

<csma> <Customer xml:lang="en">

<csma> <Name> John </Name>

<csma> </Customer>

Christian:1st pasted example above pasted by Sandro (with the And) can be used to bind to the language attribute

Christian: 2nd is used to bind to child element "Name"

Christian: And problem is when there isn't a namespace

Sandro: Even when there is a namespace, I think this syntax is odd

<sandro> sandro: ... #attribute(lang) ... seems very odd

<csma> <http://www.w3.org/XML/1998/namespace#@lang>

Christian: I have no problem with changing the syntax. The current syntax is intended to be explicit.

<csma> ?x[<Name> -> ?y]

<DaveReynolds> XML Schema component designators do use (..) in their syntax, though rather differently

<csma> ?x[<#Name> -> ?y]

<csma> ?x[<http://example.org/dummy#Name> -> ?y]

<sandro> maybe: http://www.w3.org/ns/none

Christian: But if attribute language or subelement Name have no namespace...
... in first 2 examples above don't work
... we could use a dummy namespace as above

<csma> ?x["Name"xs:NCName -> ?y]

Sandro: Dummy namespace works but it's a hack

Christian: Alternatively, we could change the spec so that an ncname is a RIF:IRI

<sandro> sandro: about about just using a string, instead of an iri?

Christian: For the frame slot name

<sandro> ?x[ ("Element", "Name") -> ?y ]

<ChrisW> time check

<sandro> ?x[ ("Element", "http://example.org", "Name") -> ?y ]

<sandro> ?x[ ("Attribute", "http://example.org", "Name") -> ?y ]

<sandro> ?x[ ("Attribute", "Name") -> ?y ]

Sandro: We could do something like above: the slot name could always be a list that would optionally include a namespace

DaveR: Lists are allowed as slot names but this would break RDF. Also..

<sandro> dave: What about XML Schema Component Designators?

<sandro> http://www.w3.org/TR/xmlschema-ref/

Christian: Good, thanks Dave, I am going to look at that document - it may provide the solution

<csma> <cust><number>1000</number></cust>

Christian: One more topic: how XML schemas are handled in RIF/XML combinations

<csma> ?x[ ("Element", "number") -> 1000 ]

<csma> ?x[ ("Element", "number") -> "1000" ]

Christian: Consider the example above: we may have a schema that tells us that cust number is an integer or we may have no schema
... if we want the interpretation to be the same in both cases, then...
... however, we may not care whether the interpretation is the same in both cases
... I think it is reasonable that the interpretation with schema would be different from that without schema, however I can understand that some people would think it's more elegant that interpretation with schema is a subset of interpretation without schema

<AdrianP> often quotes are interpreted as indicating that the value is a String instead of a numeric value

Sandro: I think all 4 cases should be handled and have a boolean flag on the import

<sandro> sandro: Let's have a boolean flag on Import where you say whether the the RIF is written using only string (the lexrep of the values) or is written to use the actual values; orthogonal to whether a schema is present.

<DaveReynolds> not convinced that using strings in place of typed values when you know the typed values makes sense

Christian: The new version of this document will be ready by the end of the week, and we need a few reviewers. Ideally, I'd like Gary to, but he's not here...will ask him by email

Sandro: I'm thinking that if you have something that's working the way you want and then you get a schema, you don't want the behavior to change

Sandro: ...and you want to use the schema for validation

<sandro> Yeah, I'm fine with schema means you have to use the datatype values, not the string.

DaveR: You can do the schema validation elsewhere. I think it's OK that if you change the semantics of the ruleset (by introducing a schema) that the behavior changes

Christian: Other volunteers for reviewing the XML data and RIF document?.... none

RIF Primer

Leora: We still need to finish the compatibility section and the test suite section
... I will work on this today

ChrisW: We have to wrap this up very soon. We can drop the sections that are not complete.

ChrisW: I can put in a paragraph on the test suite in section 9, and we can drop section 7

Leora: Does it make sense to have a RIF/RDF section without a RIF/OWL section?

ChrisW: Yes, better to have both, but one section is still useful

Christian: If we only have time for one, I think RIF/OWL is more important

ChrisW: I think RIF/RDF would be more widely used and so is more important

<Doug> agreed

DaveR: Distinction between "RIF in RDF" and "RIF and RDF"...latter is more important

<sandro> agreed -- by FAR the most important is rif AND rdf. more than rif AND owl, and more than rif IN rdf.

<csma> +1

<AdrianP> http://www.w3.org/2005/rules/wiki/Primer#RIF_in_RDF

Christian: We have already identified reviewers for RIF Primer

UCR (RIF Use Cases and Requirements)

<AdrianP> http://www.w3.org/2005/rules/wiki/UCR

Christian: We had decided to not have UCR as a current publication, but Adrian would like to revisit

AdrianP: I think it's an important document and we have already published and referenced it

<sandro> +1 rif UCR is important, if someone can do the work.

Christian: Only 2 active editors for UCR - Adrian and Leora

ChrisW: I don't think UCR is the right place for usage examples
... so, I think the document has to be either completely rewritten to be useful, or we can deprecate the document, or we can remove the examples

<sandro> +1 "there were use cases that motivated RIF"

<Doug> Maybe include the examples in an appendix as ones which led to the current state etc.

Leora: I think it's important to include in UCR use cases that cannot be repesented in any RIF dialect

Christian: As ChrisW pointed out, UCR could be a Use Case and Requirements Document as originally intended and the Primer can contain RIF examples

<Leora> leora: my point is that it's important to indicate explicitly which use cases cannot be represented in RIF dialects.

ChrisW: These are design use cases that influenced the design of the language; they are not all implementable
... I'm just saying to remove RIF syntax of the usage examples, not the examples themselves

Sandro: Don't say you can't do in RIF, rather say you need extensions to RIF to handle them

<sandro> sandro: Yes, say "you have to use an extension to do this", and make clear it cant be done in existing dialects.

Christian: Adrian and Leora will both work on updating this

<sandro> +1 for each use case, say which dialects can be used.

<csma> ACTION: Adrian to remove RIF syntax from UCR [recorded in http://www.w3.org/2010/09/07-rif-minutes.html#action03]

<trackbot> Created ACTION-1050 - Remove RIF syntax from UCR [on Adrian Paschke - due 2010-09-14].

<csma> ACTION: Leora to annotate use cases wrt implmentability w existing dialects [recorded in http://www.w3.org/2010/09/07-rif-minutes.html#action04]

<trackbot> Created ACTION-1051 - Annotate use cases wrt implmentability w existing dialects [on Leora Morgenstern - due 2010-09-14].

<AdrianP> after next week telecon, would be good to have someone for review

AOB (next meeting)

Christian: Next call next week?

<sandro> I'm at risk next week, on vacation.

<csma> next telecon will be 14 Sept

<csma> adjourned

Summary of Action Items

[NEW] ACTION: Adrian to remove RIF syntax from UCR [recorded in http://www.w3.org/2010/09/07-rif-minutes.html#action03]
[NEW] ACTION: Dave to review RIF in RDF by Sept 14 [recorded in http://www.w3.org/2010/09/07-rif-minutes.html#action02]
[NEW] ACTION: Leora to annotate use cases wrt implmentability w existing dialects [recorded in http://www.w3.org/2010/09/07-rif-minutes.html#action04]
[NEW] ACTION: Leora to review RIF in RDF by Sept 14 [recorded in http://www.w3.org/2010/09/07-rif-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2010/09/07 16:36:24 $