RDF Data Access

 

 

<bijan> no million pounds in my pocket

<bijan> Damn, works for andy but not for me :(

<AndyS> Bijan - maybe - don't type the access code to early.

<bijan> I never do

<scribe> Scribe: LeeF

<bijan> Now it just hung up on me for no reason :P)

<bijan> today, it's not going to the second bit. How weird

<SimonR> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006OctDec/0030.html

<SimonR> SimonR volunteers to scribe at the next meeting. (Oct 17?)

Next meeting October 17, Simon R to scribe

<AndyS> Minutes 3/Oct: http://lists.w3.org/Archives/Public/public-rdf-dawg/2006OctDec/0018.html

<ericP> ACTION: Bijan review rq24 against http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2006Jun/0008 [recorded in http://www.w3.org/2006/10/10-dawg-irc]

PROPOSED approve minutes http://lists.w3.org/Archives/Public/public-rdf-dawg/2006OctDec/0018.html

<ericP> DONE

RESOLVED

Action Items

<ericP> action -1

<ericP> ACTION: AndyS to edit text for DISTINCT = term-distinct [DONE] [recorded in http://www.w3.org/2006/10/10-dawg-irc]

<ericP> action -2

<AndyS> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006OctDec/0047.html

<ericP> ACTION: Bijan to propose text regarding normalization (massaging in general) while reading graphs in [CONTINUED] [recorded in http://www.w3.org/2006/10/10-dawg-irc]

<ericP> ACTION: Bijan to see if the Chilean's semantics paper offers any (??) [CONTINUED] [recorded in http://www.w3.org/2006/10/10-dawg-irc]

<ericP> ACTION: EricP to review the tests in http://lists.w3.org/Archives/Public/public-rdf-dawg/2006JulSep/0180 and say yay or nay [CONTINUED] [recorded in http://www.w3.org/2006/10/10-dawg-irc]

<ericP> ACTION: KendallC to close formsOfDistinct issue [CONTINUED] [recorded in http://www.w3.org/2006/10/10-dawg-irc]

<bijan> Actually, let's close this: [NEW] ACTION: Bijan to see if the Chilean's semantics paper offers any (??)

<ericP> ACTION: KendallC to put scope of filters at the top of next week's agenda [DONE] [recorded in http://www.w3.org/2006/10/10-dawg-irc]

<ericP> ACTION: PatH to review the proposed tests in http://lists.w3.org/Archives/Public/public-rdf-dawg/2006JulSep/0169 and say yay or nay [CONTINUED] [recorded in http://www.w3.org/2006/10/10-dawg-irc]

<bijan> Since I've done the review in some sense and we're moving into more specific stuff

<ericP> ACTION: bijan to write some text on the D-entailment issue [CONTINUED] [recorded in http://www.w3.org/2006/10/10-dawg-irc]

<ericP> action -4

<ericP> ACTION: BijanP to propose some editorial clarification text around DATATYPE [CONTINUED] [recorded in http://www.w3.org/2006/10/10-dawg-irc]

Continue discussion of scope of FILTERs

<ericP> LeeF: the apprent intention is that the scope of a FILTER comes from the nearest enclosing {}s

<ericP> ... there are related tests

<bijan> eek

<bijan> oy

<bijan> ack unmute me

<patH> bijan, you on Q?

<bijan> yes

<bijan> Until I broke it :)

<bijan> zakim that is

<ericP> ... paradox with { ... { ... ?f FILTER (!BOUND(?g) } { ... ?g FILTER (!BOUND(?f) } } will happily pass because f and g are not bound where they are filtered

<ericP> AndyS, i think scribe and i (co-scribe?) missed the boat on recording your comment

<patH> AndyS: lees design is design change, but better than current.

<bijan> oh, ok, now I'm confused

<bijan> {?x p c. OPTIONAL {?x p c.}}

<SimonR> { ... FILTER ?x } UNION { ... FILTER ?x }

<bijan> Oy

<AndyS> Typing noise?

<ericP> patH, does this work? PROPOSED: the scope of a FILTER comes from the nearest enclosing {}s and the scopes inherit from outer {}s

<ericP> welcome Fred

<patH> not sure about the 'inherit', but I think so.

<ericP> note all: no FredZ on irc

<patH> Fred, we are agenda 2, scxope of filters. Speak now or forever..

<patH> Ah, no irc, sorry.

<ericP> { { { ... FILTER ?x } UNION { ... FILTER ?x } } FILTER (whatever you want to apply to the UNION) }

<ericP> note all: FredZ now on irc

<SimonR> Do the variables in triple patterns and the variables within FILTERs act consistently (within the same group?)

bijan: I think we should decide on whether we have semantics that give the properties given in proposition 1 of SCS before deciding on the scope of filters

<SimonR> Interactions between scope and algebra, as a unifying action for the FILTER and OPTIONAL issues, et alia?

<FredZ> another interesting query: { triple OPTIONAL {...} FILTER(...) OPTIONAL {...} }

<FredZ> what is the first operand of the second OPTIONAL?

<FredZ> is it an empty pattern with a FILTER, or is it the first OPTIONAL, and the FILTER is done later

<AndyS> AndyS: blank nodes should be removed from the syntax of FILTERs if they are scoped to groups

<LeeF> fred, I think it's the first OPTIONAL

<FredZ> Lee, that's what my "SPARQL to trees" did

<LeeF> Right

<LeeF> I read through that paper, and it agreed almost entirely with my intuition and my implementation

SimonR: I strongly agree that we should settle the algebra before the scope of FILTERs

AndyS: I prefer going with a tentative design and then seeing if new information comes out when we settle the algebra

FredZ: Unhappy with the binary operators in the grammar -- FILTER and OPTIONAL -- which don't appear as binary operators in the grammar

SolutionModifier

<ericP> FredZ, does the syntax of SPARQL have this problem that you illustrated? or is it just this grammar for SPARQL?

<bijan> I wouldn't mind adding a topic on the first result of my action itme

<ericP> L := (0 1 2 3 4 5 6 7 8 9)

<ericP> LIMIT 2 OFFSET 5 gives you (5 6)

<ericP> vs OFFSET 5 LIMIT 2

AndyS: we could have a grammar that allows either order, always processed as the semantics specify now

ericP: +1 to that proposal

<bijan> LeeF, I need to chat with you about the scope thing, you have a minute afterwards?

<bijan> ROWNUMBER?!?!??!

<ericP> PROPOSED: adopt [14] SolutionModifier ::= OrderClause? OffsetLimitClause? [14a] OffsetLimitClause ::= (LIMIT (OFFSET)?) | (OFFSET (LIMIT)?) (allows "LIMIT 2 OFFSET 5" and "OFFSET 5 LIMIT 2")

<patH> shush, bijan.

<LeeF> bijan, sure

<bijan> As a presentation matter, if we had an abstract form, then the surface syntax could be more liberal

<ericP> "OFFSET 5 ORDER BY ?s LIMIT 2"

<bijan> abstain (don't care)

<FredZ> abstain

PROPOSED: adopt [14] SolutionModifier ::= OrderClause? OffsetLimitClause? [14a] OffsetLimitClause ::= (LIMIT (OFFSET)?) | (OFFSET (LIMIT)?) (allows "LIMIT 2 OFFSET 5" and "OFFSET 5 LIMIT 2")

RESOLVED, 3 abstentions

<bijan> In my case is that I don't care and I think it doesn't have any otehr effect

Is there an issue with NAF & !BOUND?

<FredZ> Eric asked if my problem with FILTER and OPTIONAL was with the grammar or with the language?

<ericP> FredZ, yes

<FredZ> The answer is that I have struggled to rewrite the grammar to make FILTER and OPTIONAL into conventional binary operators

<AndyS> There is a proposed chnage for optional - Fred - I've not had feedback on that.

<FredZ> and I have not found a way.

<ericP> aha. language then

<FredZ> Andy: yes, your idea is a step forward, though ideally i'd like to see

<FredZ> a BNF like: OptionalPattern ::= GroupGraphPattern OPTIONAL GroupGraphPattern

<LeeF> example of NAF via OPTIONAL and !bound: http://thefigtrees.net/lee/sw/sparql-faq#universal

<ericP> patH, most popular use case i remember was: find all the foaf:Persons who are missing an mbox

<LeeF> actually, the example at the URL above is of MIN

<bijan> I lost everything from andy

<ericP> http://www.w3.org/TR/2004/WD-rdf-sparql-query-20041012/#unsaid

<bijan> I lost that

<bijan> What?

<Zakim> AndyS, you wanted to ask if there is new information (UNSAID)

<AndyS> UNSAID is a filter despite the appearance of a graph pattern

<bijan> Uh, is that link from ericP supposed to resolve to:

<bijan> Section status: working group is not working on this feature at the moment. It is currently likely to be dropped from the SPARQL query language.

<bijan> ?

bijan, yes

<ericP> yea, as i said, it just says that it's at risk, and we later backed that up by removing it

<SimonR> The use of NAF is based on a closed world assumption. Can we approach giving people NAF a better way. For example, in some way allowing people to explicitly say "this graph knows everything about X, so NAF applies in this particular case"

<AndyS> http://www.w3.org/2001/sw/DataAccess/issues#unsaid

<bijan> +1

<ericP> +1 to LeeF

<ericP> cwm has log:notIncludes but it always get used on the command line with another invocation (which just moves the expressivity into /bin/sh)

<SimonR> Just for the record, my observation is that trying to do this with a single negation operator is always going to be confusing; the modal approach is the only approach I can think of that is fairly simple.

<bijan> http://lists.mindswap.org/pipermail/pellet-users/2006-September/000884.html

<bijan> If it's punted, it's punted

<patH> yes, its punted until someone has new information to offer.

Continue discussion on open world and other value tests

PROPOSED to adjourn

ADJOURNED.