-
DRAFT -
RIF Telecon 4 Sept 07
4
Sep 2007
See
also: IRC log
Attendees
Present
csma, Dave_Reynolds, AllenGinsberg,
josb, LeoraMorgenstern, Harold, +43.512.507.9aaaa, Stella_Mitchell, ChrisW,
barry_b, PaulVincent, DougL, Sandro, Gary_Hallmark, DavidHirtle, luis_polo,
Hassan_Ait-Kaci, Michael_Kifer, DaveReynolds
Regrets
Chair
Christian de Sainte-Marie
Scribe
Allen Ginsberg
Contents
<ChrisW> Scribe: Allen Ginsberg
<ChrisW> scribenick: AllenGinsberg
<csma> Chair: Christian de Sainte Marie
<csma> Scribe: Allen Ginsberg
<csma> scribenick: AllenGinsberg
Admin
csma: action review
chrisW: action 334 closed,
335 done
minutes accepted for aug 28 telecon
Liason
csma: no action to review
<PaulVincent> OMG PRR: no news
F2F
csma: everyone please
fill out f2f survey
<sandro> F2F7 Register/Regrets
chrisW: no unpdates on f2f
csma: first draft agenda
by end of next week
... skipping UCR item because Axel not here
... any volunteer to review another one?
<Harold> As I mentioned, Axel and Paula told me they
have only a shakey connection from the Reasoning Web Summer School in
chrisW: any volunteers for
next week?
csma: Axel should do it
next week
<ChrisW> http://www.w3.org/2005/rules/wg/wiki/D._Evaluation%3A_Use_Cases
chrisW: dave you did use
case 8?
dave: stuff has changed
<ChrisW> http://www.w3.org/2005/rules/wg/wiki/UC8_Worked_Example
dave: can do UC 8 next
week
AllenG: can do UC3
following week
csma: move on to BLD
UCR
BLD
- RDF
csma: action review
... sandro action 336?
sandro: continued
csma: 337?
harold: continued
csma: 338
chrisW: 338 is done (by
Gary)
<Harold> New actions come in for me, too, with the
discussion and work Sandro started for the successor to asn06.
<ChrisW> Harold, do you want to record some actions?
csma: jos to discuss
changes to RDF compatibility section
<josb> http://www.w3.org/2005/rules/wg/wiki/Core/RIF-RDF_Compatibility
<Harold> No, that's an 'informal' action.
jos: cleaned it
up....make more readable and to incorporate new items and resolutions
... also included new comments and discussion itmes
... biggest thing, found more elegant way to define the semantics, so now
easier to read etc
csma: does anyone have
clarification qustions?
chrisW: I have one...in
common interpretations section condition 2
... what is condition 2 doing?
jos: just makes sure
that whenever you use an element as a property then it is in the set Ip
AllenG: (technical
discussion of condition 2)
chrisW: just says all slots
are RDF properties?
jos: yes
... Michael had some skepticism concerning the semnatics?
... especially concerning the combination semantics; I responded in email
mkifer: I didn't get a
chance to read it yet.
jos: let's continue in
email
csma: did you (Jos) mean
that an RDF graph can be translated into RIF rules?
jos: yes.
... actually translated to facts
csma: that means rif
includes rdf?
jos: no...it means that
if you want to use RDF with RIF you are implictly assuming this semantics
csma: does it mean that
any RDF can be expressed as a RIF rule set?
jos: any kind of rdf
entailment maps into entailment in rif
mkifer: this is the
problem: the combined semantics isn't needed because rif already allows for
that embdding
<josb> http://lists.w3.org/Archives/Public/public-rif-wg/2007Sep/0001.html
jos: let's continue that
over email
mkifer: ok
csma: let;'s go over
other issues in the email one-by-one
jos: 3 issues related to
identifiers...
... absolute iri's versus rdf use of uri references
... might need a conversion?
dave: rdf uses "rdf
uri references"
... they were trying to hone in on iri's before the irs spec was finalized
... i suggest we just leave it as iri's
<sandro> DaveReynolds: What the RDF spec uses is
"RDF URI References" which is a confusing term. It's not "URI
References". It was, rather, their best attempt to anticipate what IRIs
would be. There may be a difference around spaces. I suggest we not dwell on
any differences.
jos: I was confused by
different statements in the specs
<sandro> Jos: I was trying to figure out if we need to
take into account the conversions between URIs and IRIs.
dave: we don't need to
worry about conversion
<sandro> DaveReynolds: I don't think we do. We just
treat them as IRIs.
<sandro> +1 (just treat them as IRIs)
dave: in the iri specs
conversion algorithms would be defined
... we certainly wouldn't define new conversions ourselves
csma: dave we are missing
some of your audio....
<DaveReynolds> yes
csma: so the solution is
to use absoluete iris
chrisW: do we need to refer
to uris at all?
<DaveReynolds> exactly, don't point to any conversion,
just talk about IRIs
jos: for the sake of rdf
compatibility
sandro: rdf uri references
are not uris
<josb> http://www.w3.org/TR/rdf-concepts/#section-Graph-URIref
chrisW: they were iri's
before iri's were defined
sandro: I agree we dave
that we don't need to worry about these issues
... we are basically tracking an evolving standard
<DaveReynolds> I proposed a form of words (derived from
the SPARQL spec) before, I can find that again
sandro: just talk about
iri's and maybe include a note to deal with special cases
jos: investigate
further?
sandro: not us.
csma: do we need a formal
decision?
... any objections to just talking about iri's?
none
<sandro> Sandro: The issue that I think is real is what
we say you do about the odd little corner cases like an RDF URI Reference that
is not an IRI -- eg one with a space in it.... I suggest we ignore this issue.
(I think concurring with Dave on this suggestion)
jos: 2nd issue: some rdf
literals are not strings accroding to xml schema strings
... but we didn't decide whether we want to go with xmls 1.0 or 1.1
<Harold> We certainly should be prepared for 1.1 and
for now encourage RIF users to stay in the common subset.
jos: not clear about all
the consequences of doing one or the other
chrisW: can we leave these
decision to implementors of rif?
csma: do you mean in rif
instance docs?
jos: yes we could, but
then we should say something about rdf literals that are not xml 1.0 strings
chrisW: what does xmls 1.0
say?
<sandro> Jos: we need to say what to do if you see an
xml 1.0 literal string
jos: doesn;t say
anything
dave: we should leave the
value space unconstrained (as in xmls 1.1)
... that doesn't stop people from using 1.0 libraries to process stuff
... in the spec we need to point to a specific spec
... 1.1 isn't at spec yet.
csma: but we never
investigated the consequenses of using 1.1 as reference for xml types
dave: agreed...there may
be other issues, e.g., builtins
csma: somebody to take an
action with regard to xml datatypes included in rif
... volunteers/
... we need someone with good xml expertise
... gary?
gary: haven't been
tracking 1.0 vs 1.1
<ChrisW> important point jos made, we can't just leave
this up to implementors because we agreed to include XML datatypes directly as
part of RIF syntax
dave: what about asking
the xml schema group?
csma: yes
<DaveReynolds> not me
jos: i can
... but they haven't responded to earlier email yet
sandro: sending a comment
to working group would be good
... i can try talking to the people directly, or send a message to working
group
csma: actions?
<ChrisW> ACTION: debruij to send message to
XML Schema WG comments list [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action01]
<rifbot> Sorry, couldn't find user - debruij
action taken by jos to send comment to xml schema working group
<ChrisW> ACTION: jdebruij to send message to
XML Schema WG comments list [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action02]
<rifbot> Created ACTION-339 - Send message to XML
Schema WG comments list [on Jos de Bruijn - due 2007-09-11].
jos: ill-typed literals
in rif vs rdf
... treatment is different
<josb> http://www.w3.org/2005/rif/rdf-ill-typed-literal/uri-encode("s"^^u)
jos: I made a suggestion
(see link in irc)
... map ill-typed literals to uris. Is this a good idea?
<sandro> rifbot, status
csma: don't ill-typed
literals in rules make them uninterpretable and therfore who cares?
jos: they can occur in
rdf data, so we need a way to treat them in rif
csma: ah this is for the
combination semantics?
dave: this seems to be an
unimportant case. in actual data you don't let ill-typed literals through. this
isn't practically important
... just say the embedding only works for well-typed case
csma: basically agree with
dave
jos: i don't agree. why
not support all of rdf?
csma: it doesn't mean
that rif rules can't interact with all of rdf data, it only means that some rdf
data can't be translated into rif rues
<Harold> Jos' proposal re ill-typed literals could
still come into RIF: where we deal with exceptions, partial compliance, etc.,
anyway.
<Harold> (see Sandro's earlier wiki page)
jos: but this comes into
play with queries - rif query over rdf-data (containing ill-typed literals)
csma: but if rdf knows
what to do with them, then rdf can deal with them
jos: reiterates point
about query-answering
chrisW: something that
understands rdf can do the translation for rif vs. rif itself knows what to do
... how complicated is this? if uncomplicated we can handle all of rdf
csma: but at what cost?
jos: only
implementations that care about rdf would be affected
csma: but this
contradicts your combination semantics argument,
... when you try to embed the rdf graph as rif rules you would get a syntax
erro (with ill-typed literals)
chrisW: in support of
csma's argument:
<sandro> Jos: I DONT propose we support ill-typed
literals, instead I propose we translate to URIs.
jos: rif doesn't need to
support ill-typed literals, we are translating to uri's
chrisw: so you are really
in agreement (with csma)
<josb> http://www.w3.org/2005/rif/rdf-ill-typed-literal/uri-encode("s"^^u)
sandro: what do the uri's
look like?
jos: shows link in irc
csma: ok
<MichaelKifer> should the data type name be part of
such a uri?
jos: does this comply
with best practices?
sandro: this is a pretty
weird thing to do (but legal, i think)
... it;s a weird hack, but might be our best option
csma: any objections to
doing this?
mkifer: basically a good
idea, but need to work on details, e.g., other languages might want to use this
mechanism
sandro: why did rdf do
this? probably because it would be impossible to decide whether the literal is
really ill-typed
csma: so we table this
issue for now?
sandro: jos can you edit
the page to show an actual example/
jos: yes
<josb> http://lists.w3.org/Archives/Public/public-rif-wg/2007Sep/0013.html
jos: there are 4 additional
issues (see link in irc)
csma: once we resolve all
these issues are we done with compability?
jos: it depends...doc
would still need examples, etc. There are other related issues, but we are
pretty close
csma: do we need to raise
those other issues formally?
... yes
<ChrisW> ACTION: chris to investigate raising
Jos' identified issues [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action03]
<rifbot> Created ACTION-340 - Investigate raising Jos\'
identified issues [on Chris Menzel - due 2007-09-11].
csma: chris will you ask
deborah to create those seven issues?
chrisW: ok
... sandro can you remove chris menzel from the list?
sandro: yes
<ChrisW> ACTION: cwelty to investigate raising
Jos' identified issues [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action04]
<rifbot> Created ACTION-341 - Investigate raising Jos\'
identified issues [on Christopher Welty - due 2007-09-11].
<ChrisW> zakim close item 5
Arch
- Data Models
csma: action review
action 331 continued
action 330 done
csma: action 258 done?
dave: yes a long time ago
... we discussed 258 at the last f2f.
csma: action 258 closed
action 256 continued
<scribe> ACTION: 254 to semantics section for
arch doc [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action05]
mkifer: at least telecon i
mentioned splitting the arch doc into two.
csma: let's obsolete this
action for now
action 254 closed
<DaveReynolds> http://www.w3.org/2005/rules/wg/wiki/Arch/Data_Sets
csma: dave do you want to
discuss your changes
dave: see link in irc.
... this is a result of discussion at last f2f.
... there are now 3 subsections relating to differnt isues teased out
... how to identify the data set used by rules
... basically convey the metadata to give an identifier for data set
... 2nd section: data model identification
... how do yo tell the rule process what data model is being used?
... at f2f we agreed on supporting 3 data models
... i updated metadata vocabulary to accomodate those three
... 3rd section data-model usage: see email discussion about this
csma: so let;s use
remaining time to how to embed metadata in rif
jos: there is no
mechanism for specifying meta data in rif
... it's straigtforward to do it, but we need to decide what we want to include
csma: all we have is the
proposal to use rdf to express metadata
dave: right.
Allen: I missed some of
this...
<Harold> Dave, the current syntax proposal is in http://www.w3.org/2005/rules/wg/wiki/Core.
sandro: the question is
related to extensibility.
... if extensibility is easy then we can let people write their own metadata
schema
<ChrisW> who is cgi-irc?
csma: we need to get a
better feeling of what kind of metadata would be required
<cgi-irc> apparently me for some reason... strange
<cgi-irc> Doug L
csma: if we only a small
set of metadata maybe we can just get away with afew attributes and values
chrisW: this is orgthogonal
to whther metadata is done in rdf
<ChrisW> -1 to extenc
<DaveReynolds> don't mind
<Hassan> -1 to extend
<sandro> adjourn
meeting adjourned
<DaveReynolds> bye
Summary of Action Items
[NEW] ACTION: 254 to semantics section for
arch doc [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action05]
[NEW] ACTION: chris to investigate raising
Jos' identified issues [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action03]
[NEW] ACTION: cwelty to investigate raising
Jos' identified issues [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action04]
[NEW] ACTION: debruij to send message to
XML Schema WG comments list [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action01]
[NEW] ACTION: jdebruij to send message to
XML Schema WG comments list [recorded in http://www.w3.org/2007/09/04-rif-minutes.html#action02]
[End of minutes]
$Date: 2007/09/04 16:31:40 $