SV_MEETING_TITLE -- 16 Feb 2010

<AxelPolleres> oh, indeed...

<kasei> I can if necessary, but the agenda is SD-heavy, so might be talking a lot..

<AxelPolleres> greg... let's do the following, you start, when SD comes on, you just take over chairng that item, and I scribe, if ivan isn't here yet

<kasei> Axel, sure.

<AxelPolleres> scribe: Greg Williams

<AxelPolleres> scribenick: kasei

<LukeWM> Garlik has LukeWM, SteveH

<AxelPolleres> Zakim, who is on the phone?

<LukeWM> zakim who is on the phone?

<ivan> scribe: Ivan

<Souri> ??p34 is me

<scribe> scribenick: ivan

admin issues

<AxelPolleres> PROPOSED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2010-02-09

AxelPolleres: anybody comment?
...
...
...
...

<AxelPolleres> RESOLVED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2010-02-09

once... twice...

next meeting

in a week

AxelPolleres: we have to think about f2f agenda
... i am already discussing with lee
... main issue is to go towards last call
... that is also look at the issues list
... if editors can look at their priorities

comments

<AxelPolleres> http://www.w3.org/2009/sparql/wiki/Comments

AxelPolleres: we have all open comments assigned last time
... aggregate is on the agenda for later today
... anybody has any news to report?

once... twice... ... ...

AxelPolleres: none

liaisons

rdb2rdf, egov, and rif

AxelPolleres: anything to report?
... sandro, would you mind to take over the liaison to rif?

sandro: is fine

AxelPolleres: any issues you would be aware of?

sandro: none

AxelPolleres: rdb2rdf?

Souri: nothing major, we did two groups now
... we are working on two sets of documents and then on use cases, nothing related to sparql

AxelPolleres: SteveH anything for egov?

SteveH: I cannot really see to participate on that group, would be good to have another liaison

sandro: that would be me :-(

AxelPolleres: would you be able to report occasionally?

sandro: I will try...

summary on sd issues

<AxelPolleres> topiv: SD issues

<AxelPolleres> http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JanMar/0245.html

kasei: the recent comment and my response yesterday sums up the state
... the first thing is the available graph issue

<kasei> http://kasei.us/2009/09/sparql/sd-example.ttl

kasei: i have been keeping up to date on the comments on that
... the modeling of the available graphs is in the ??? descriptions
... I think lee's suggestion is the right way to move forwards
... modulo the name
... we would have graph collections
... resources would bring together all available the graph collection

AxelPolleres: you said a dataset is a subclass of collection

kasei: it remains any number of named graphs, but it also contains a default graph

AxelPolleres: the question was whether we would advertise different datasets?

kasei: we talked about that earlier and we agreed that it would not make sense to define more datasets

AxelPolleres: ok

<AxelPolleres> ack ??P32

<LukeWM> ??P32 is us

<LukeWM> we shouldn't be on the queue

AxelPolleres: what is the concrete issue on that particular question

<LukeWM> ack ??P32

kasei: i do not believe there was any open issues on that
... will check

AxelPolleres: what is next on your list

kasei: property functions issue
... i have heard arguments on both sides

<AxelPolleres> next issue... property functions declarations

kasei: whether we would support an explicit name property function that a service supports
... i would be fine including this
... lee expressed concerns
... wording would be tricky, there is no spec we can refer to
... it is legal under sparql but is not discussed anywhere

AxelPolleres: we have not even had this under the elected features

<SteveH> +1

<SteveH> not +1

kasei: it seems that it exists in the gray area like the default graph being the union of all named graphs, is there and not explicitly discussed in the spec
... i am willing to include it unless somebody objects and modulo a proper wording

SteveH: i would object to it
... it is bizarre to standardize
... the example of default graph is explicitly in the spec, so this is not the same

kasei: i am not sure they are all that difference on the level of specing in the documents

<AxelPolleres> some reservations against saying something about property functions...

SteveH: a section in the spec that explicitly does that
... I forget exactly which

AxelPolleres: we have a dataset description out of our scope, we refer to VoiD
... would not this a possible candidate for extension like VoiD?
... although we do not have any external spec

<SteveH> kasei, section 8.1 "RDF data can be combined by the RDF merge [RDF-MT] of graphs. One possible arrangement of graphs in an RDF Dataset is to have the default graph be the RDF merge of some or all of the information in the named graphs."

kasei: the difference that VoiD fits nicely to the way we model things
... here we would have a property referring to a URI describing the function
... but we lack the spec

AxelPolleres: there is a difference in how it works, but both have an ontology describing that thing
... if we have a link describing a dataset, couldn't have a link to something describing the property function
... i am not 100% sure

kasei: i would be hesitant to merge those two
... if two datasets are implemented by two services, one having a function and the other not, you do not want to describe that as part of the dataset description, this is an implementation issue

AxelPolleres: you would like to have it, and SteveH does not, we may want to open an issue and solve that at some point

SteveH: it seems very peculiar for a rec on a feature that is not being standardized

<AxelPolleres> ISSUE: Do we need (descriptions of) property functions in SD? Is this in scope for us?

<trackbot> Created ISSUE-54 - Do we need (descriptions of) property functions in SD? Is this in scope for us? ; please complete additional details at http://www.w3.org/2009/sparql/track/issues/54/edit .

kasei: part of the motivation that one can describe implementations that are convergent
... many people have this feature and is widely implemented
... my support for this is very mild at this point
... lee's suggestion is that it is so widely implemented and it requires very little for our part adding it

SteveH: to be useful we have to standardize the actual function;

kasei: you can do that in the extension functions already

orri: you cannot refer to a standard name that you cannot define
... it would be very useful if this behaviour did have a name
... even standardizing what we mean would not ask for lots of space in the document

<pgearon> I'd like to see a description that listed predicates that provide property functions

<AxelPolleres> link to the email?

kasei: we could say that if you see that property function it would act differently

AxelPolleres: any more new insight at this point?

<pgearon> +1 to what kasei said

AxelPolleres: or put it on email

<kasei> http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JanMar/0316.html

<chimezie> I would be interested in seeing how the current spec can be said to provide a mechanism for this

<chimezie> because I don't see how it would

AxelPolleres: I would suggest that people who care about this responds
... is there anything else?

<kasei> having trouble hearing over typing

kasei: the other issues are not controversial just have to be done

<kasei> still need to talk about issue 43?

AxelPolleres: one more comment, there was one issue on entailment
... whether we can different entailment on different graphs

bglimm: AndyS wanted to say that one graph uses rdfs and the other one simple, for example

AxelPolleres: that is issue 43, which is still open

<kasei> scribenick: kasei

<ivan> scribenick: ivan

<AxelPolleres> ISSUE-43 ... if we need to have different entailments described, we need aproposal

subtopic: powder

<AxelPolleres> greg?

AxelPolleres: any powder expertise

kasei: this was to see if we can include a way to talk about a collection of graphs
... and powder might be able to do this

<AxelPolleres> could anyone jump in here with help on POWDER?

ivan: if somebody asks the question on that, I am happy to try to answer it

<AxelPolleres> ACTION: greg to get back to ivanh on the usabuility of POWDER for SD. [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action01]

<trackbot> Created ACTION-189 - Get back to ivanh on the usabuility of POWDER for SD. [on Gregory Williams - due 2010-02-23].

alternative aggregate design, comment from della valle

<AxelPolleres> http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2010Feb/0006.html

AxelPolleres: we had an action on SteveH
... is it too early to discuss it?

SteveH: I do not feel to have a good grasp on what it does
... it seems to be just syntax

AxelPolleres: i read through it and wrote a mail to the list with open questions
... mostly clarification questions
... i would like to get back to to emmanuele
... in general there are merits in the proposal
... the main benefit seems to be that it allows to define several aggregates on the same solution set in parallel
... i am not sure we can do that
... we would need to have several subqueries

SteveH: you cannot have different groupings on aggregate functions
... in the current syntax

AxelPolleres: he had an example where he can do something more concise than we cat

orri: sql has a model with different grouping for aggregates

AxelPolleres: orri, could you look at that mail as well to see how this is addressed

orri: on the sql side we can do this rollup side, and it would be quite possible to take the sql semantics to sparql

AxelPolleres: is it in our scope

SteveH: it is in our scope, i am not sure about the syntax
... you have to know a lot to understand what is happening
... in its current form it is not very tempting

AxelPolleres: it would be interesting to see how it is done in sql
... Orri, can you get back on that?

Orri: I will look it up

<AxelPolleres> ACTION: Orri to send some comment to the list how multiple groupings at once work in SQL [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action02]

<trackbot> Created ACTION-190 - Send some comment to the list how multiple groupings at once work in SQL [on Orri Erling - due 2010-02-23].

AxelPolleres: in the meantime, if there are no objections, i will get back to emmanuelle for clarification questions though that would not be the official answer

test cases

AxelPolleres: there is some wiki page
... not too much at the moment

<AxelPolleres> http://www.w3.org/2009/sparql/wiki/index.php?title=Category:Testing&action=edit

AxelPolleres: the main point is to define test cases
... the mail of emmanuelle brings in some interesting test cases
... the question is who would do it, what is the way to proceed

<AxelPolleres> ... we have some in the spec?

AxelPolleres: opinions?
... any experience how the test cases were elaborated in the last group?

(more enthusiasm required!!!)

<Souri> Lee would be the best person to do that

AxelPolleres: we can pick some victims to extract the test cases in the form of examples
... and elaborate the test cases
... we could do that for query, update, etc
... what we need is basically volounteers

(Axel desperately trying to generate enthusiasm...)

AxelPolleres: if nobody comes forward I can make some starting points and pick victims
... if nothing else works

(everybody ducks...)

AxelPolleres: if I look at the features than we have aggregate functions, update, service descriptions, and of course entailment and property paths
... and query

<AxelPolleres> * Query language expressivity

<AxelPolleres> o Aggregate functions

<AxelPolleres> o Subqueries

<AxelPolleres> o Negation

<AxelPolleres> o Project expressions

<AxelPolleres> * Update

<AxelPolleres> * Service description

<AxelPolleres> Entailment RDFS

<AxelPolleres> Entailment OWL

<AxelPolleres> PropertyPaths

AxelPolleres: it would be a good starting point
... any volounteers for queries and aggregate functions?
... I might go ahead I might do this

:-(

<AxelPolleres> ACTION: Axel to collect test cases for aggegate functions [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action03]

<trackbot> Created ACTION-191 - Collect test cases for aggegate functions [on Axel Polleres - due 2010-02-23].

<dcharbon2> I volunteer for subqueries

<OlivierCorby> I am ok 2

<AxelPolleres> ACTION: David to collect testcases for subqueries [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action04]

<trackbot> Sorry, amibiguous username (more than one match) - David

<trackbot> Try using a different identifier, such as family name or username (eg. dcharbon2, dnewman2)

<scribe> ACTION: dcharbon2 to collect test cases for subqueries [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action05]

<trackbot> Created ACTION-192 - Collect test cases for subqueries [on David Charboneau - due 2010-02-23].

<AxelPolleres> ACTION: Olivier to collect testcases for subqueries [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action06]

<trackbot> Created ACTION-193 - Collect testcases for subqueries [on Olivier Corby - due 2010-02-23].

<AxelPolleres> ACTION: chime to collect testcases for negation [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action07]

<trackbot> Created ACTION-194 - Collect testcases for negation [on Chimezie Ogbuji - due 2010-02-23].

<AxelPolleres> ACTION: Olivier to collect test cases for Project Expressions. [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action08]

<trackbot> Created ACTION-195 - Collect test cases for Project Expressions. [on Olivier Corby - due 2010-02-23].

<AxelPolleres> ACTION: Axel to collect test cases for Project Expressions. [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action09]

<trackbot> Created ACTION-196 - Collect test cases for Project Expressions. [on Axel Polleres - due 2010-02-23].

SteveH: the format does not work for update

<pgearon> +q

AxelPolleres: so that has to be updated for update indeed

<AxelPolleres> Update... we probably need some format extension for update.

pgearon: i can work to redraft the framework for that

<AxelPolleres> ... test cases for update absolutely needs adaption

pgearon: i can work with somebody to ensure that it would be all right

<AxelPolleres> ACTION: Paul to get back to Andy to work on how to adapt the test cases format for update. [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action10]

<trackbot> Created ACTION-197 - Get back to Andy to work on how to adapt the test cases format for update. [on Paul Gearon - due 2010-02-23].

<OlivierCorby> ok for property path

<AxelPolleres> ACTION: Olivier to collect test cases for property paths [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action11]

<trackbot> Created ACTION-198 - Collect test cases for property paths [on Olivier Corby - due 2010-02-23].

(AxelPolleres has succeeded in creating enthusiasm!!! Kudos:-)

<sandro> Top of the hour, gotta go.

<AxelPolleres> ACTION: Birte to collect test cases for entailment [recorded in http://www.w3.org/2010/02/16-sparql-minutes.html#action12]

<trackbot> Created ACTION-199 - Collect test cases for entailment [on Birte Glimm - due 2010-02-23].

<LukeWM> bye

<bglimm> Bye.

<kasei> Axel: can you close action 176 for me?

<AxelPolleres> adjourned

<AxelPolleres> greg, sure.

- DRAFT -

SV_MEETING_TITLE

16 Feb 2010

Attendees

Contents