SV_MEETING_TITLE -- 16 Mar 2011

rrrsagent, make records public

<yvesr> Zakim: ??P4 is yvesr

<scribe> scribe: mbrunati

<JFB> +34.67.92.aaaa is JFB

<cygri> no zakim

<cygri> oh hi zakim

thanks

<LeeF> sandro, it's already setup

<LeeF> it's "rdfwg"

<sandro> LeeF, it was not....

<LeeF> yes it was :)

<sandro> Lee, then explain "--> Zakim (rrs-bridgg@128.30.52.169) has joined #rdf-wg"

who is talking?

<LeeF> ah well, i'd have to first explain "[11:01] * Zakim (rrs-bridgg@128.30.52.169) Quit (Client exited)" i guess :)

<sandro> LeeF, ahhhhhh

<gavin> Oh dear

zakim goes down

twice

<AndyS> and trackbot?

<gavin> trackbot does seem to be here

<davidwood> Minutes from last week: http://www.w3.org/2011/rdf-wg/meeting/2011-03-09

minutes last telecon

<pfps> minutes look fine to me

<davidwood> Propose to accept

<tomayac> +1

<NickH> stupid americans ;-)

<davidwood> ▪ Action item review:

<davidwood> http://www.w3.org/2011/rdf-wg/track/actions/open

ivan: dst time and one hour before for the Europeans

<pfps> Action 17 appears to be done

<trackbot> Sorry, couldn't find user - 17

<davidwood> ACTION-17: Gavin's response at http://www.w3.org/2011/rdf-wg/wiki/GraphConceptTerminology

<trackbot> ACTION-17 Try and produce a digram based on the g-box, g-snap, g-text model from Sandro's email and this conversation notes added

<cygri> gavin, very nice diagram!

gavin has made the diagram on GRAPH

<AndyS> gavin, "me:" and ":me" are the same?

action-6

<gavin> Andy, yes... thanks. Will fix :D

<davidwood> F2F1

<davidwood> http://www.w3.org/2011/rdf-wg/wiki/F2F1

action-items, one more week with an hour earlier for the Europeans

now f2f

please express the intent to make presence explicit on the f2f on the wiki

there are some action-items for the next week

action-4 and action6 are overdue

<Zakim> sandro, you wanted to ask about hotels

more work on the GRAPH Task force

<NickH> sandro: yvesr and I are staying at Mercure Hotel Amsterdam Aan de Amstel

<yvesr> we're staying at the Mercure hotel

<yvesr> (near Amstel station)

<NickH> I believe SteveH and mischat are too

sandro: on hotels on the f2f at the Mercure hotel

GRAPH TF, a proposal

<davidwood> Graph task force: Richard's proposal: http://www.w3.org/2011/rdf-wg/wiki/TF-Graphs/RDF-Datasets-Proposal

a proposal to make something concrete to work on

how to address better the use cases, take the concept of RDF core, SPARQL

RDF dataset identify in SPARQL are basically syntactic

they are based on something already defined

the discussion on the ML is on two points

<gavin> In addition to the named graphs and default graph mentioned by Richard many SPARQL implementations support a Union graph as well.

semantics and SPARQL: is it complete or not?

what the relations between graph, bnodes, what identify graph, g-snaps or snapshot and so on

<AxelPolleres> well, SPARQL will say *something* about semantics in its Entailment Regimes document, but there will not be interdependencies between graphs in Datasets in SPARQL Entailment regimes.

<AndyS> Axel, isn't entailment is only for graphs?

another point is maybe this proposal is something that is complete or not?

<Zakim> manu, you wanted to discuss data-deduplication in RDF as it relates to bnodes.

manu:

what are the proper directions on using bnodes in this context? g-snaps

<gavin> +q to mention Union graph and it's usefulness

thinking on hard drives, the core OS thinking on files on different ways, and bnodes

we can think on them on the same way

bnode as a file container maybe... we really need bnodes as a cross g-snaps

bnodes are probably a problem for SPARQL users

<sandro> davidwood: The problem I see with bnodes is when they are returned from a query and people want to follow-up.

<gavin> Nope.

<Zakim> AndyS, you wanted to talk about graph/subgraph

andy:

<sandro> AndyS: sometimes you know one graph is a strict mathematical subgraph, eg a inference view of the data and the base data, then you know it's the same bnode.

<manu> Manu: Looking at this from a data de-duplication perspective, perhaps bnodes could be viewed as only being an implementation concern?

richard:

<sandro> +1 test cases on bnodes

<sandro> +1 test cases on bnode sharing

<yvesr> +1

<manu> Manu: That is, you can share the bnode data across multiple instances in a copy-on-write way... that would allow us to say that bnodes are only valid inside a particular g-snap (it would make our decision easier)

sandro:

<manu> Manu: So, one could expect thousands of bnodes that are exactly the same between multiple g-snaps... but that storage space could be compressed using bnode de-duplication on disk.

<davidwood> g-handle? :P

people use graphs and not are talking about graphs...

<manu> Manu: However, there are other concerns as David and Andy mention about queries on bnodes.

what's the relation between URIS and g-snaps

<AndyS> SPARQL's "named graph" is the pair without meaning of the relationship -- maybe we need a "best practices" for using that

<gavin> +1 sandro

URIs as g-boxes

<manu> +1 to sandro's terminology

<FabGandon> +1 for URIs identifying g-boxes

<manu> Graph Box -> Graph Snapshot -> Graph Representation

<manu> Graph Resource -> Graph Snapshot -> Graph Representation ?

<AndyS> +1 to g-snap of a g-box (the IRI) at some point in time

we need to determine more terminology

gavin:

<gavin> http://www.w3.org/TR/sparql11-http-rdf-update/

<manu> Maps nicely to Resource -> Snapshot -> Representation - perhaps we should add this to TAG/WEBARCH people?

on the rdf update on the SPARQL WG

what is the mapping? in turtle

<AndyS> New name of that doc is "SPARQL 1.1 RDF Dataset HTTP Protocol"

gavin: other thing in the document, SPARQL mentions union graph

<Zakim> gavin, you wanted to mention Union graph and it's usefulness

what graph are coming from, what is dataset notion in this context

richard:

<manu> +1 sharing blank nodes between two g-snaps matters with union graphs.

union graph, URIs are g-boxes

if we want to store some metadata about the graph,

we have to identify a sort of snapshot of that graph

<danbri_> [could someone ask Guus if he could let me into his office?]

with or without some trasformation on it?

richard:

<sandro> sandro: Excellent example, fetching http://www.w3.org and putting the metadata into a dataset, you'd have to do a (reversable) syntactic transformation, eg into http://myserve.example.com/fetched=http://www.w3.org

how to understand how to move forward on it?

sandro has pointed out more attention on terminology

<cygri> propose that a g-sandro is a (URI, g-snap) pair

<gavin> Yeah, but SPARQL already uses it?

<yvesr> cygri: :)

<gavin> FROM NAMED

what are graphs, definitions, better named graph definitions maybe, some confusion on defs

<sandro> a "name g-box pair" ?

URI, IRIs graph, we have to more focus on named graph on SPARQL

how to get out metadata about and out of this graph?

andys:

URI as id for the graph or maybe

much meaning in a word, in a graph, in a name, few weeks to make clearer it

richard wants to make a more practical nextstep on it

more work on the ML

<sandro> sandro: It would be great to get the wording around "named graph" made somehow clearer in SPARQL drafts before LC, or at least before PR.

david: now Turtle

<davidwood> http://www.w3.org/2011/rdf-wg/wiki/TF-Turtle/Proposals

richard:

<AndyS> was Richard again (+1 for all the effort)

thanks

<davidwood> Yes, +1 to Richard

name graph, more work, some proposal are out of scope

more usable, more use cases, many different languages, special characters

some proposals on graph TF

<gavin> +1 as it's also great for building parsers ;)

david:

<AndyS> Can we not start with the n-triples text in RDF test cases?

a comment on Twitter that strongly not Turtle in RDF rec

we have to understand what is out of scope on the f2f

it's important to make ideas freely in this WG, even if someone upset and so on

<manu> I don't think that we need the "!" syntax in TURTLE, either... our company has never had a need for it.

<gavin> At least for Turtle task force it doesn't look like very many people are going be at the F2F :(

f2f on the JSON proposals, it's important to move on also

<manu> +1 to get the preliminary JSON document sorted by the end of 1st F2F

david: what the timeline of closing out the opened issues?

we not have an opened issue on named graph

<sandro> issue: What is a named graph and what should we call it?

<trackbot> Created ISSUE-14 - What is a named graph and what should we call it? ; please complete additional details at http://www.w3.org/2011/rdf-wg/track/issues/14/edit .

david: are there other opened issues on this telecon?

<sandro> issue: what is the relationship between the IRI and the triples in a dataset/quad-syntax/etc

<trackbot> Created ISSUE-15 - What is the relationship between the IRI and the triples in a dataset/quad-syntax/etc ; please complete additional details at http://www.w3.org/2011/rdf-wg/track/issues/15/edit .

<gavin> other people have been having trouble

david: how to understand better what kind of documents we supposed are working on?

we have to use Turtle or not?

maybe it is necessary to collect on wikis periodical revisions on materials discussed

JSON TF now

manu:

<manu> syntax options for JSON: http://www.w3.org/2011/rdf-wg/wiki/JSON_Syntax_Options

<tomayac> +1 for nathan's work!

<manu> http://www.w3.org/2011/rdf-wg/wiki/TF-JSON#RDF_in_JSON_Use_Cases

<davidwood> Kudos to the JSON TF for the longest wiki page :)

<manu> http://www.w3.org/2011/rdf-wg/wiki/TF-JSON#RDF_in_JSON_Design_Requirements

collected some info on JSON syntax options, use cases, design requirements

<manu> http://www.w3.org/2011/rdf-wg/wiki/JSON_Syntax_Options

+1 for nathan's work

<manu> http://www.w3.org/2011/rdf-wg/wiki/JSON_Syntax_Options#Syntax_Structure

the primary decision is

JSON based on triples or objects model

or document approach

to move forward we have to make some decision, but it's not clear how

maybe we have missed out some information on these proposals

<Zakim> LeeF, you wanted to ask if we're putting the cart before the horse

leef: probably we need to know what problems we try to solve

<AndyS> +1 to LeeF

<manu> the object-based vs. triple-based decision hints at the problem we're trying to solve.

<cygri> +1 to LeeF

+1 manu

<sandro> sandro: "make existing JSON map to RDF model" -- "make RDF easy for JSON users" -- "make json serialization of RDF"

<yvesr> indeed - the answer will depend on whether we wanta full rdf serialisation or not

<manu> http://www.w3.org/2011/rdf-wg/wiki/TF-JSON#A_primary_goal_SHOULD_be_to_build_a_human-friendly_version_of_the_serialization_for_JSON_developers

<sandro> (actually, some of that was Lee)

tomayac:

<AndyS> +1 to conflates issue with implementation

<yvesr> +1

<sandro> +1 at-a-glance serializations.

<NickH> +1

maybe we not need a complete serialization on json

not now

thanks

<sandro> ACTION tomayac to create and start to populate a wiki area of demo graphs with how they each look in each json proposal.

<trackbot> Created ACTION-20 - Create and start to populate a wiki area of demo graphs with how they each look in each json proposal. [on Thomas Steiner - due 2011-03-23].

cygri: more documentation on use cases and so on...

triple vs object approach

the problem maybe is on the head of us, how to manage a triplestore from a javascript context

<yvesr> although SPARQL results in JSON is probably not in our scope

<sandro> cygri: some people find it important to, eg, access a triplestore without a (turtle) library. or, "i have an RDF graph and want to write it in json". or manu's thing I don't quite understand.

<sandro> cygri: is the motivation to add bits of annotaiton to existing json, so they can be parsed as RDF?

<sandro> cygri: or do you not need migration/annotation

manage the existing json stuff or to generate RDF to JSON syntax

<Zakim> manu, you wanted to respond to Richard.

or something else?

manu:

<sandro> manu: (1) annotation pre-existing json, and (2) take data people are emitting today (eg twitter) and understand that as RDF, by adding a "default context"

shall we make a json complete serialization of RDF?

cygri:

<sandro> +1 manu --- don't worry about another rdf serializaion, but DO both make it easy for json developers, AND try to including existing json data via annotations

<davidwood> Scribe: mbrunati

the discussion is on how to manage graphs, and all RDF stuff on JSON context,

<sandro> cygri: This goal of trying to parse RDF out of existing json may be out of scope for this WG.

<Zakim> AndyS, you wanted to ask about the relationship to RDFa-WG RDF API (process Q)

david: not enough time to complete the agenda on RDF Rec probably

<Zakim> manu, you wanted to propose a separate RDF in JSON call to work through these issues.

AndyS: relations between RDF - WG and RDFa

<tomayac> wondering whether the any-json-2-rdf-json wouldn't be a one-off mapping each single time. can there be a generalized approach?

RDFa-WG and WG RDF API correct

<sandro> manu: The RDF WebApps WG will do RDFa API and RDF API.

manu: there is a strong relation between these WG, the point is how to accomplish it?

<sandro> +1 manu, make a doodle poll for a telecon time

<davidwood> Sandro, do you wish to respond to Manu or address setting up a call for JSON?

manu: we need a separate call to make clear this misunderstanding probably

sandro:

it's not clear but it take months to manage resources on it

david:

<manu> ACTION: Manu to create a doodle poll to find a time to have a call. [recorded in http://www.w3.org/2011/03/16-rdf-wg-minutes.html#action01]

<trackbot> Created ACTION-21 - Create a doodle poll to find a time to have a call. [on Manu Sporny - due 2011-03-23].

probably time is finished at this time

<gavin> Regrets for the next meeting. And was the scribe. Jeremy will be here instead!

thanks

<tomayac> bye

<pchampin> bye

<LeeF> cheers, everyone

<hsbauer> bye

<sandro> gavin: I can't scribe next week, but jeremy will

<danbri_> posted draft FOAF use case for named graphs in http://www.w3.org/2011/rdf-wg/track/actions/4

<danbri_> not sure exactly what was wanted so mostly historical, with sketch of requirements

- DRAFT -

SV_MEETING_TITLE

16 Mar 2011

Attendees

Contents

GRAPH TF, a proposal

Summary of Action Items

Scribe.perl diagnostic output