IRC log of rdf-wg on 2011-04-13

Timestamps are in UTC.

07:56:20 [RRSAgent]
RRSAgent has joined #rdf-wg
07:56:20 [RRSAgent]
logging to
07:56:25 [gavinc]
Mmm... 1 am is in fact morning I guess ;)
07:56:27 [ivan]
rrsagent, set log public
07:56:44 [ivan]
gavinc: just a minute, we will dial in soonish
07:56:50 [Zakim]
+ +1.404.978.aabb
07:57:19 [tomayac]
zakim, aabb is me
07:57:19 [Zakim]
+tomayac; got it
07:57:35 [Guus]
Guus has joined #rdf-wg
07:59:54 [OlivierCorby]
OlivierCorby has joined #rdf-wg
08:01:56 [pgroth]
pgroth has joined #rdf-wg
08:03:13 [davidwood]
davidwood has joined #rdf-wg
08:05:31 [Zakim]
+ +31.20.592.aacc
08:05:54 [pfps]
zakim, who is here?
08:05:54 [Zakim]
On the phone I see gavinc, tomayac, +31.20.592.aacc
08:05:55 [sandro]
zakim, who is on the call?
08:05:56 [Zakim]
On the phone I see gavinc, tomayac, +31.20.592.aacc
08:06:11 [AZ]
AZ has joined #rdf-wg
08:06:35 [AZ]
08:06:50 [pfps]
08:06:51 [sandro]
Ivan Herman
08:06:54 [sandro]
08:06:57 [sandro]
Dan Brickley
08:07:03 [sandro]
08:07:10 [sandro]
Paul Groth (guest)
08:07:15 [sandro]
08:07:23 [sandro]
Jan W.
08:07:30 [sandro]
Jean Paul Inria
08:07:34 [sandro]
Nick Humphrer
08:07:37 [sandro]
08:07:44 [davidwood]
davidwood has joined #rdf-wg
08:07:47 [sandro]
08:07:47 [Zakim]
+ +
08:07:57 [sandro]
08:08:00 [sandro]
08:08:05 [sandro]
Steve Harris
08:08:05 [OlivierCorby]
zakim, aadd is me
08:08:05 [Zakim]
+OlivierCorby; got it
08:08:10 [ivan]
08:08:19 [sandro]
08:08:21 [yvesr]
08:08:28 [sandro]
Sandro Hawke
08:08:33 [sandro]
David Wood
08:08:36 [sandro]
08:09:13 [Zakim]
+ +1.760.705.aaee
08:09:24 [AZ]
zakim, +1.760.705.aaee is me
08:09:24 [Zakim]
+AZ; got it
08:09:28 [ivan]
zakim, who is here?
08:09:28 [Zakim]
On the phone I see gavinc, tomayac, +31.20.592.aacc, OlivierCorby, AZ
08:09:47 [sandro]
AZ, you're a bit loud
08:09:55 [AZ]
zakim, mute me
08:09:55 [Zakim]
AZ should now be muted
08:10:01 [ivan]
zakim, aacc is ivan
08:10:01 [Zakim]
+ivan; got it
08:10:03 [pfps]
zakim, aacc is CWI
08:10:03 [Zakim]
sorry, pfps, I do not recognize a party named 'aacc'
08:10:25 [sandro]
zakim, ivan is Meeting_Room
08:10:25 [Zakim]
+Meeting_Room; got it
08:10:53 [SteveH]
Scribe: mischat
08:11:00 [mischat]
08:11:27 [FabGandon]
zakim, who is here?
08:11:27 [Zakim]
On the phone I see gavinc, tomayac, Meeting_Room, OlivierCorby, AZ (muted)
08:11:55 [mischat]
Guus: are we happy with the agenda ?
08:12:11 [mischat]
Guus: does anything need to be amended?
08:12:46 [mischat]
so thomas is not here so Marco (?!?) will be giving the json roundup
08:12:54 [mischat]
08:13:20 [tomayac]
08:13:38 [mischat]
08:17:27 [raphael]
raphael has joined #rdf-wg
08:18:23 [mischat]
is everyone physically at CWI turning up to dinner tonight ?
08:18:26 [mischat]
if not shout ...
08:18:54 [mbrunati]
mbrunati has joined #rdf-wg
08:19:02 [mischat]
anyone for agenda changes ?
08:19:16 [mischat]
we are looking at this now
08:19:16 [mischat]
08:19:26 [mischat]
^^ are the objectives for this f2f
08:19:57 [mischat]
Guus: this f2f to move us from an open discussion to a more targeted effort
08:20:24 [mischat]
we are looking to get documents in place
08:21:02 [mischat]
from now on we should have our long threads turn into something tangible and useful for the process
08:21:33 [mischat]
we are now looking to identify starting documents for the various tasks
08:21:50 [mischat]
Guus: would like to have names against the various documents, so that we can push work forward
08:21:59 [mischat]
Graph's tasks force
08:22:40 [mischat]
we have some standard terminology now in terms of GraphTerminology
08:23:11 [mischat]
Guus: another issues is the alignment with the SPARQL work
08:23:26 [NickH]
08:24:03 [mischat]
Guus: so what will be the starting document for the GRaphs TF, should it be the RDF concepts
08:24:05 [mischat]
08:24:19 [mischat]
that is the current feeling, and these are things which we need to discuss
08:24:38 [tomayac]
thanks, NickH for the photo :-)
08:25:18 [mischat]
Guus: we have some cleanup tasks, and there are discussions needed to identify what changes need to happen to the various RDF documents
08:25:48 [mischat]
Guus: we seem to have a good grasp of the issues, re: a good issue list has been developed
08:25:59 [mischat]
Guus: do people think we have a good grasp of the problem domain ?
08:26:04 [mischat]
question for the room ^^
08:26:30 [FabGandon]
for ecah identifier we define (e.g. g-box identifiers) we should also discuss what happens when we dereference that identifier (e.g. what do I get when I dereference the IRI of g-box? triples in the g-box? triples about g-box? both)
08:27:57 [pgroth]
- moving on to discussing turtle
08:28:03 [FabGandon]
Guus: for TURTLE starting point is the doc from team submission
08:28:56 [FabGandon]
Guss: N-triple considered as a limited sub-set of Turtle
08:29:04 [mischat]
mischat has joined #rdf-wg
08:29:07 [ivan]
zakim, who is here?
08:29:07 [Zakim]
On the phone I see gavinc, tomayac, Meeting_Room, OlivierCorby, AZ (muted)
08:29:22 [NickH]
pchampin: very impressive!
08:30:16 [ivan]
zakim, Meeting_Room has David Wood, Sandro, Mateo, Steve Harris, Fabien, Pierre Antoine, Cygri, Yves, Nick, Jean-François, Jan, PFPS, Paul Groth, Chris Matheus, Dan Brickley, Misha Tuffield, Ivan
08:30:16 [Zakim]
+David, Wood, Sandro, Mateo, Steve, Harris, Fabien, Pierre, Antoine, Cygri, Yves, Nick, Jean-François, Jan, PFPS, Paul, Groth, Chris, Matheus, Dan, Brickley, Misha, Tuffield,
08:30:20 [Zakim]
... Ivan; got it
08:30:35 [mischat]
JSON, we have documents to start from, in terms of the Talis submission
08:30:47 [yvesr]
if we're not able to standardise object-based json, can we at least standardise a canonical mapping from an rdf graph to some straight-fw json?
08:31:14 [mischat]
daveWood: just asked about Talis submitting a member submission
08:31:48 [mischat]
Guus: it is important to figure out what is achievable in terms of work in the JSON TF
08:32:31 [mischat]
danbri: JSON developers learn new formats all the time
08:33:04 [mischat]
danbri: we can get it wrong, and push out three syntaxes, and we will get it right eventually
08:33:47 [mischat]
in the JSON TF, we need to elicit what our objectives should be
08:34:07 [mischat]
if we develop more than one syntax then we will have doubled the work
08:34:18 [mischat]
ivan: asked about cleanup related actions
08:34:26 [mischat]
Guus: there is time set aside for that tomorrow
08:34:51 [mischat]
Guus: has no idea how much work the cleanup will be
08:35:35 [mischat]
ivan there are a bunch of small issues, URIRef vs IRI
08:35:42 [pchampin]
ivan: following discussion on the ML, we need to agree on what 'deprecation' means for this WG
08:36:00 [tomayac]
(audio no longer understandable on the US no. anyone else on the phone have this issue, too?)
08:36:04 [mischat]
ivan: the meta-issue regarding "deprecation" should be discussed and sorted out here at the f2f
08:36:33 [mischat]
the issue will be tackled tomorrow, but we are going to try and touch upon it now
08:36:34 [mischat]
for 20 mins
08:37:12 [ivan]
08:37:17 [mischat]
so lunch at somepoint between 12:30-13:00 central european summer time
08:37:46 [mischat]
davewood: re: turtle, dave wants to know what standardised will be developed by tthe WG
08:37:56 [tomayac]
(audio back to normal. phew)
08:38:26 [Danbri]
Danbri has joined #rdf-wg
08:38:27 [mischat]
i.e. we will have turtle, will we have qturtle, trig, or what combination of serialisations will we develop
08:39:12 [tomayac]
sandro FTW! thanks!
08:39:15 [mischat]
peter: question should we have Qturtle, or turtle, should one be a superset ?
08:39:37 [mischat]
so dave would like to see issue sorted out
08:40:00 [mischat]
SteveH: said we could have one document which lists all of the turtle(related) serialisations
08:40:16 [pchampin]
sounds like a great idea to me
08:40:31 [mischat]
Dave's goal for the f2f is to nail the turtle work
08:40:48 [mischat]
so we have clear goals, turtle work seems to be the most advanced
08:41:27 [mischat]
danbri, we have a big archive "www rdf comments", will someone go through the archives
08:41:38 [mischat]
where we have had lots of feedback from people about RDF
08:42:27 [tomayac]
sandro, i up-scale it client-side, works perfect for me. thanks!
08:42:31 [davidwood]
davidwood has joined #rdf-wg
08:43:24 [davidwood] - historical RDF issues
08:43:35 [mischat]
we are going to move on to the Graph's discussion, if we are happy with the objectives ?
08:43:59 [mischat]
Richard is about to give some slides summarising the graphs work
08:44:28 [danbri]
danbri has joined #rdf-wg
08:44:52 [mischat]
there are some slides on the wiki for richard's talk
08:44:56 [cygri]
08:45:48 [mischat]
cygri: will talk about the "problem", the "open issues", and will give us a view of other potential issues
08:46:17 [mischat]
the charter says we must standardising a model for multiple graphs
08:47:06 [mischat]
the charter also states that we must standardise turtle and a something similar with multi graph support
08:47:34 [mischat]
a decision was made for the turtle to focus on syntax and the graphs tf can look at extending turtle
08:48:16 [pfps]
It's *turqle*!!!
08:48:21 [mischat]
davidwood: missed the call where the work of putting in mutlli graph support to turtle should be a task for the graphs tf
08:49:01 [mischat]
Guus: turtle tf can talk about the syntax, but the graphs tf will inform what the multi graph syntax should represent
08:49:21 [mischat]
cygri: is listing inputs to the graphs tf
08:49:37 [mischat]
sparql's rdf dataset: ( and sparql update's graph store)
08:49:42 [mischat]
being one
08:49:50 [mischat]
Carroll et al " Named Graphs
08:49:56 [mischat]
Notation3: quoted graphs
08:50:36 [FabGandon]
I disagree with the idea that "named graphs" in RDF/XML should be only "if time permits", for me it's a must
08:50:49 [mischat]
n3 allows for nesting, and quoting graphs, the n3 work should definitely inform the named graphs tf
08:51:10 [ivan]
FabGandon: any modification to RDF/XML is time permitting
08:51:15 [mischat]
Trig, and Nquads should help inform any syntax discussions
08:51:48 [mischat]
cygri: also FabGandon has request to add named graph support to RDF/XML (like trix)
08:52:06 [mischat]
Reification was mentioned as an input
08:52:14 [mischat]
and finally Typed graph literals
08:53:05 [mischat]
cygri: is pointing to a wiki page which has the named graph use-cases
08:53:13 [sandro]
david: Can we just view Reificiation as a way to address named graphs, and once we do that, we can more cleanly deprecate reification?
08:53:48 [mischat]
which is broken down into the following : 5 storage use cases, 2 query use cases, 8 provenance, 4 use for standard foundation for w3c specs, 2 advanced annotations use case
08:54:17 [mischat]
cygri: stated how we dont seem to be using the use-case we have in many of the discussion
08:54:25 [mischat]
we have lots of use-cases, they should be used
08:54:33 [mischat]
we have a bunch of proposals in this space
08:55:06 [mischat]
we have 2 concrete proposal in this space so far
08:55:10 [tomayac]
sandro, small is good enough for me.
08:55:25 [mischat]
cygri: there are implied proposals
08:55:50 [mischat]
i.e. that n3's style quoted graphs may be more useful than the RDF dataset stuff
08:56:07 [mischat]
cygri: is walking through the issues
08:56:15 [mischat]
issue-5 : graph literals
08:56:32 [mischat]
issue-5 asks where we should have graph literals
08:56:57 [mischat]
08:57:16 [mischat]
issue-14 : what is a named graph and what should we call it ?
08:57:43 [mischat]
these include : Named Graph, named g-box ?, g-pair, or even IRI-graph-binding
08:58:27 [mischat]
ivan: would have liked to have seen a slide on "g-*" syntax
08:58:47 [mischat]
so that we can have agreement on what the terms are
08:59:26 [gavinc]
depends on how we quoted it ;)
08:59:56 [mischat]
Guus: we need to come up with decent names for the g-* terminology, Guus personal opinion is that we need to make sure we dont use the overloaded term "graph" without qualifying it
09:00:18 [mischat]
we need to make sure that we all agree on what the various g-* terminology is
09:01:17 [mischat]
pgroth: said that Luc Moreau Provenance WG has given feedback on the g-* syntax
09:01:24 [mischat]
see mischat's email to the list ^^
09:01:38 [mischat]
issue-15 : "g-pair" semantics
09:01:53 [mischat]
we have a couple of options re: this issue
09:02:06 [mischat]
1: Leave it undefined (abstract syntax only)
09:02:13 [Zakim]
+ +1.408.642.aaff
09:02:14 [mischat]
2: or we could define it
09:02:25 [davidwood]
09:02:44 [davidwood]
09:02:59 [zwu2]
zwu2 has joined #rdf-wg
09:02:59 [yvesr]
cygri: "is a person a g-box?"
09:03:06 [mischat]
there are issues there re: scoping the terminologies used
09:03:10 [zwu2]
zakim, mute me
09:03:10 [Zakim]
sorry, zwu2, I do not know which phone connection belongs to you
09:03:50 [zwu2]
zakim, +408.642.aaff is zwu2
09:03:50 [Zakim]
sorry, zwu2, I do not recognize a party named '+408.642.aaff'
09:03:57 [Zakim]
09:04:04 [ivan]
zakim, aaff is zwu2
09:04:04 [Zakim]
+zwu2; got it
09:04:10 [zwu2]
thanks ivan
09:04:14 [mischat]
issue-17: graph merging
09:04:14 [trackbot]
ISSUE-17 How are RDF datasets to be merged? notes added
09:04:18 [zwu2]
zakim, mute me
09:04:18 [Zakim]
zwu2 should now be muted
09:04:20 [yvesr]
in n3, there is a property in between the graph and the IRI, which makes that relationship explicit
09:04:24 [mischat]
there are issues re: blank nodes and merging
09:04:34 [mischat]
and what would happen when merging graph datasets
09:05:20 [mischat]
Guus: thinks that the main issue with extending RDF Semantics will be re: RDF merge, and posed as a question to peter
09:05:47 [mischat]
peter doesn't know what exactly what is needed, sparql has a notion of graph merge
09:06:00 [mischat]
ivan: we are informally bound by what sparql does
09:06:21 [Danbri]
Danbri has joined #rdf-wg
09:06:25 [mischat]
Guus: we should make sure that sparql and rdf align
09:06:47 [mischat]
issue-21 : sharing Node IDs
09:07:06 [mischat]
nodeId being bnode identifer
09:08:12 [mischat]
cygri: the issue talks about the same bnode identifier in a quad based a trig file, how are the bnodes to be scoped ?
09:10:14 [mischat]
davidwood: thinks that we are going to be making strong statements about scoping bnodes and pushing it up to the RDF standards, but we should make sure that what we do doesn't break implementations
09:11:23 [mischat]
issue-22 (empty graph)
09:11:53 [mischat]
the issue is asking what we should be doing in terms of multi-graph support and empty graphs
09:12:05 [mischat]
trig, nquads, and sparql all do something different
09:12:13 [mischat]
issue-23 (multigraph media types)
09:13:00 [mischat]
the issue asks whether we should change mime-types if we add graphs to existing serialisations
09:13:16 [Danbri]
q+ to ask (no rush) re graph literal datatypes, whether a media types-as-Uris would be better than just defining our own for rdf syntaxes
09:13:43 [mischat]
issues: discussion volume : Graph Literals was the most talked about issue in the named graph tf
09:14:00 [mischat]
davidwood: asked about consensus re: graph literals
09:14:47 [mischat]
cygri: candidate issues : Do we need nesting of graphs ?
09:15:16 [mischat]
what is "nesting of graphs" ?
09:15:22 [mischat]
could we have an example
09:15:42 [mischat]
cygri: thinks that is would be hard to do without the graph literals
09:15:56 [mischat]
ivan: essentially this is a syntax issue
09:17:04 [mischat]
in the nested graph, or graph literals dont need to have a named graph
09:17:59 [mischat]
we are about to create a new issue
09:18:32 [mischat]
cygri: we don't know that the question is right now
09:19:07 [Zakim]
09:19:10 [mischat]
yvesr: states we need to have use-cases for the "nesting of graphs"
09:19:37 [mischat]
danbri: wonders whether it is a syntax question
09:20:17 [sandro]
ISSUE: Do we need syntactic nesting of graphs (g-texts) as in N3?
09:20:18 [trackbot]
Created ISSUE-28 - Do we need syntactic nesting of graphs (g-texts) as in N3? ; please complete additional details at .
09:20:25 [mischat]
SteveH: thinks that it would be a syntax issue only if graphs with nesting could be serialised into some non-nesting serialisation such as turtle
09:21:15 [mischat]
davidwood: an open-issue re: how do we refer to graphs
09:21:38 [mischat]
cygri: asked how do you name a graph
09:21:40 [Danbri]
Of course we could nest multiple-graphs too ("here are the quads I downloaded from .... Yesterday")
09:22:07 [mischat]
cygri: goes back to issue-15 and asks whether that covers dave's issue
09:22:43 [mischat]
cygri: next proposed issue, do we need a "default graph" ?
09:23:02 [mischat]
do we need to align with sparql, but we definitely need to define what a default graph is
09:23:33 [Danbri]
(default graph for The Web? :)
09:24:20 [mischat]
davidwood: believes that AndyS's point re: "default graph" is that we should not be throwing away early thinking in terms of allowing people to define their own notion of default graph
09:25:19 [mischat]
Guus: two important alignment issues with SPARQL, how do RDF datasets related to g-boxes and more specifically what is the relation between SPARQL's default graph and default graphs in RDF
09:26:00 [davidwood]
davidwood has joined #rdf-wg
09:26:06 [mischat]
Guus and cygri would like an issue with alignment default graph from sparql
09:26:24 [mischat]
peter would argue against the default graph
09:26:49 [davidwood]
09:26:56 [davidwood]
ack Danbri
09:26:56 [Zakim]
Danbri, you wanted to ask (no rush) re graph literal datatypes, whether a media types-as-Uris would be better than just defining our own for rdf syntaxes
09:27:14 [mischat]
danbri: what would count towards to qualifying a triplestore dump in terms of default graph
09:27:41 [sandro]
steve: the SPARQL WG has backed itself into a corner wrt defaults.
09:27:56 [sandro]
pfps: give it a name, but throw the name away when you're done
09:27:57 [mischat]
SteveH: says that the sparql group doesn't have a set resolution for this stuff
09:28:35 [pgroth]
hey sandro, after the end of this discussion am I allowed to raise issues as an observer?
09:28:40 [pgroth]
or anybody
09:28:43 [Danbri]
09:28:48 [mischat]
cygri thinks there should be a relation between sparql's dataset, default graph
09:29:00 [sandro]
ISSUE: Do we support SPARQL's notion of "default graph"?
09:29:00 [trackbot]
Created ISSUE-29 - Do we support SPARQL's notion of "default graph"? ; please complete additional details at .
09:29:26 [sandro]
ISSUE: How does SPARQL's notion of RDF dataset relate our notion of multiple graphs?
09:29:27 [trackbot]
Created ISSUE-30 - How does SPARQL's notion of RDF dataset relate our notion of multiple graphs? ; please complete additional details at .
09:29:46 [Danbri]
Davidwood, zakim had an earlier q queued from me re graph literals - happy to defer if this is wrong point for it
09:29:47 [mischat]
two separate issues : 'Do we support SPARQL's notion of "default graph"', and how does 'How does SPARQL's notion of RDF dataset relate our notion of multiple graphs'?
09:30:02 [mischat]
cygri: asks do we NEED a concrete syntax for multi-graphs
09:30:41 [mischat]
cygri: says that the charter talks about lots of syntax related work, does this need to be pushed upstream and do we need to standardise this concrete syntax
09:30:47 [Zakim]
09:30:58 [davidwood]
Danbri: Please cover that when the list of candidate issues has been cleared, but before we move onto a new topic.
09:31:24 [Danbri]
09:31:27 [Danbri]
09:31:29 [Danbri]
09:31:46 [sandro]
ISSUE: Do we produce a standard (REC) syntax for conveying multiple graphs?
09:31:46 [trackbot]
Created ISSUE-31 - Do we produce a standard (REC) syntax for conveying multiple graphs? ; please complete additional details at .
09:31:52 [mischat]
SteveH: asks where should this work be standardised
09:32:23 [danbri_]
danbri_ has joined #rdf-wg
09:32:54 [mischat]
cygri: asks whether the potential Reification deprecation should live in the cleanup tasks, or should it be in the graph's TF
09:33:42 [mischat]
ivan: and sandro think that the reification cleanup will be scoped out properly depending on the outcomes of the graphs tf
09:34:00 [mischat]
cygri: now lists the minimal work to get to what the charter states
09:34:26 [mischat]
1. Lift SPARQL's RDF Dataset into RDF Concepts and Abstract Syntax
09:34:48 [mischat]
2. Evaluate additional possible features based on use cases
09:35:02 [mischat]
3 Do not define a concrete syntax
09:35:18 [mischat]
4 If we MUST have a concrete syntax standardize N-Quads
09:35:33 [mischat]
5 Avoid multigraphs in RDF/XML, JSON, Turtle, and rdfa
09:35:42 [pgroth]
I would like to raise the following three issues, if I'm allowed:
09:36:08 [pgroth]
1) Can g-snaps be identified?
09:36:19 [zwu2]
like the N-quad idea
09:37:13 [pgroth]
2) can the working group define which kinds of graphs are considered a resource
09:38:11 [mischat]
danbri: has a question re: graph literals, maintenance, and what you would have to do. Would you require to mint a new URI for each media-type to support graph literals, danbri wonders whether we would just be recreating the mime-type registry
09:39:11 [danbri]
q+ to ask what form of advice we ought to be offering to
09:39:17 [pfps]
Scribe: pfps
09:39:57 [pfps]
Guus: can we work in the issue list?
09:40:00 [sandro]
09:40:36 [pfps]
pgroth: what about identification of all the various g-strings?
09:40:41 [sandro]
pgroth: Can g-snaps be identifies or just g-boxes?
09:40:43 [FabGandon]
q+ to talk about concrete syntax and use cases and RDF/XML
09:40:53 [mischat]
pgroth: is talking about this issue, which i forwared to the list :
09:41:39 [pfps]
cygri: this depends on the relationship between an IRI and the "graph"
09:41:46 [danbri]
(re URIs for mediatypes, see prev discussion )
09:41:53 [sandro]
cygri: graph literal is one way to do that. another is that maybe with named graphs is iris identifiy g-snaos. another is immutable g-boxes.
09:41:58 [pfps]
cygri: does the IRI refer to the g-box or the g-snap, or whatever
09:42:38 [pfps]
pgroth: Provenance WG happy to defer to the RDF WG for a solution, but we want something
09:42:49 [mischat]
sorry pchampin 1 sec
09:42:51 [sandro]
ISSUE: Can we identify both g-boxes and g-snaps?
09:42:51 [trackbot]
Created ISSUE-32 - Can we identify both g-boxes and g-snaps? ; please complete additional details at .
09:43:09 [pfps]
danbri: what about using timestamps to fix the value of a changing g-box
09:43:23 [mischat]
q+ on the provenance WG
09:43:44 [pfps]
pgroth: Provenance needs a language for the provenance of resources
09:44:02 [ww]
suggestion: uuid for fixing value of changing g-box rather than timestamp
09:44:10 [mischat]
pchampin: i sent an email to the list today, my mail headers claim this is the URI, but it 404's for me too : <-- sorry
09:45:10 [danbri]
09:45:35 [pchampin]
pchampin: not sure I understand what "provenance of a resource" means...
09:46:15 [Zakim]
09:46:18 [pfps]
scribe note: provenance of resources -> provenance of resources that are graphs
09:46:37 [mischat] <-- provenance WG charter
09:46:54 [ww]
provenance of a document makes sense... provenance of a resource is harder to pin down i think
09:47:27 [ww]
a g-snap being a certain kind of resource more like a document where it also makes sense...
09:47:43 [pfps]
pgroth: provenance graphs can be relative to a particular viewpoint - which might involve part of a particular g-snap
09:47:46 [Gendor]
Gendor has joined #rdf-wg
09:48:04 [ww]
provenance of the resource that is my cup of coffee is more complicated and probably out of scope
09:48:06 [sandro]
pgroth: Is there a way to select and refer to a subset of a g-snap?
09:48:16 [sandro]
mischat: ... or individual triples.
09:48:28 [pfps]
mishat: also from provenance - want to talk about particular triples
09:48:52 [ww]
to talk about a particular triple is to talk about a graph of size 1, no?
09:48:56 [pfps]
davidwood: provenance issues can result in very many graphs (e.g., hundreds of thousands)
09:49:01 [ww]
or is there a salient difference?
09:49:14 [danbri]
eg. has per-triple annotation in a graph API
09:49:19 [pfps]
pgroth: yes, e.g., creating a named graph for each triple
09:50:07 [sandro]
ISSUE: Do we provide a way to refer to sub-graphs and/or individual triples?
09:50:07 [trackbot]
Created ISSUE-33 - Do we provide a way to refer to sub-graphs and/or individual triples? ; please complete additional details at .
09:50:26 [ww]
need some kind of inhertance - don't need to materialise graphs for each triple, just imply them, and they inherit the provenance information that makes sense from their super-graph
09:50:36 [pfps]
danbri: some (many?) graph stores allow access to things like individual triples (as graphs)
09:51:41 [danbri]
the example I give is ...they have written adaptors for a number of graph stores-
09:51:53 [pfps]
pgroth: does the WG need an issue about individual triples as graphs, etc.
09:52:06 [pfps]
guus: let's wait until we determine whether it is needed
09:52:34 [pfps]
mischat: there are many other related issues, like signatures
09:52:39 [pfps]
ivan: signatures are out of scope
09:53:01 [pfps]
mischat: what about ordering of triples in a graph
09:53:21 [pfps]
ivan: syntax may provide an answer
09:53:23 [danbri]
graph stores that have per-edge annotation:
09:53:49 [ww]
if signatures were in scope, defining an ordering to compute the signature would make sense, but generally there is no ordering, right?
09:54:06 [pfps]
sandro: the SPARQL construct can (and often does) create small graphs, including individual triples
09:54:07 [danbri]
rrsagent, pointer?
09:54:07 [RRSAgent]
09:54:15 [sandro]
agreed, ww
09:54:31 [Guus]
09:54:48 [Guus]
09:54:52 [pfps]
davidwood: we may need to worry about distinguishing between the various g-* when naming
09:54:57 [mischat]
09:55:08 [pfps]
zakim, who is here?
09:55:08 [Zakim]
On the phone I see gavinc, tomayac, Meeting_Room, OlivierCorby, AZ (muted), zwu2 (muted), OlivierCorby.a, OlivierCorby.aa,, OlivierCorby.aaaa
09:55:11 [Zakim]
Meeting_Room has David, Wood, Sandro, Mateo, Steve, Harris, Fabien, Pierre, Antoine, Cygri, Yves, Nick, Jean-François, Jan, PFPS, Paul, Groth, Chris, Matheus, Dan, Brickley,
09:55:13 [Zakim]
... Misha, Tuffield, Ivan
09:55:21 [ww]
per-edge annotation: actually the annotation is the predicate i think. two nodes make an edge (s,o), and the predicate labels the edge
09:55:49 [ww]
maybe the graph is a second lable for the edge
09:55:59 [danbri]
ivan, has ' The proposal for the group has now been accepted and the group operates under its final charter' but that link 404s
09:56:23 [ivan]
danbri, reload
09:56:39 [danbri]
ack danbri
09:56:39 [Zakim]
danbri, you wanted to ask what form of advice we ought to be offering to
09:56:39 [pfps]
pchampin: SPARQL construct is a g-text (sort of)
09:57:04 [pchampin]
s/construct is a/construct returns a/
09:57:10 [sandro]
davidwood: If you do a GET on an IRI and get a gtext, isnt that IRI naming a g-box? Well, if that IRI happens to be a SPARLQ-end-point plus SPARQL Construct Query, then you've just given a URI to a subgraph....
09:57:28 [pfps]
danbri: what about RDF Web Applications group - they will make an API for RDF - what is the relationship to this WG?
09:57:42 [ww]
davidwood: what about the same sparql operation with POST?
09:57:49 [pfps]
cygri: This hasn't been discussed yet
09:58:03 [sandro]
09:58:17 [pchampin]
@david: I have no problem with considering http://../sparql@construct... as identifying a g-box
09:58:36 [pfps]
ivan: the API may be just a simple as "IRIs can be used to retrieve a graph"
09:58:54 [pfps]
ivan: RDFa has no syntactic sugar for named graphs, and probably won't go there
09:59:50 [pfps]
danbri: does this WG need to provide something to the RDF Applications group
09:59:56 [pfps]
ivan: not necessarily
10:00:29 [FabGandon]
ack FabGandon
10:00:29 [Zakim]
FabGandon, you wanted to talk about concrete syntax and use cases and RDF/XML
10:00:42 [pfps]
fabien: three questions
10:00:54 [pfps]
fabien: 1/ I want a concrete syntax - for provenance,
10:01:14 [pfps]
fabien: 2/ in many applications we use RDF/XML so we want named graphs in there
10:01:44 [danbri]
Guus/Davidwood, cygri ... I guess implicitly we resolve something like "this group does not believe it has specific items to deliver around RDF-Graph that impact the ability of the new RDF Web apps API group to make progress"
10:02:11 [pfps]
guus: at Shanghai there was discussion on this, which lead to changes to the charter
10:02:51 [pfps]
ivan: this WG can decide whether (or not) to touch RDF/XML (probably to create a new, superset)
10:02:59 [danbri]
RDFAPI charter = "RDF API, Recommendation: This document will define a generic API for managing RDF data. "
10:03:02 [pfps]
ivan: I don't know whether this is needed
10:03:11 [pfps]
guus: this might become a general issue
10:03:57 [pfps]
cygri: issue 23 talks to this, at least in a general sense
10:04:38 [pfps]
ivan: there might be other changes for RDF/XML, e.g., a schema-friendly version
10:05:18 [pfps]
ivan: I am afraid that changing RDF/XML would end up being a lot of effort
10:05:33 [danbri]
10:05:48 [pfps]
guus: we have to consider these issues
10:06:26 [zwu2]
fabien, can n-quad satisfy your provenance requirements?
10:06:54 [danbri]
Re XML -- we've had 13 or so years for the community to come up with a more beautiful XML notation for RDF. Nothing has emerged. Does anyone really think attempting that work in committee would improve things?
10:06:57 [pfps]
fabien: 3/ link to SPARQL construct - which produces RDF/XML, so augmenting RDF/XML might involve a link to the SPARQL WG
10:07:10 [danbri]
closest attempt "You can think of this syntax as Notation 2. A later syntax, Notation 3, was much more successful."
10:07:18 [pfps]
cygri: I don't think that there would be a link here
10:07:43 [pfps]
fabien: this might argue against extending RDF/XML
10:08:01 [gavinc]
The original named graph paper Jeremy Carroll, et al... had a method of describing named graphs in RDF/XML
10:08:12 [danbri]
+ we had a *whole wg* creating GRDDL to map from idiomatic XML into RDF (anyone using GRRDL?)
10:08:14 [pfps]
davidwood: if we want to change RDF/XML we need XML experts, and there are lots of other things that would end up on the table
10:08:43 [pfps]
ivan: there are also no proposals for any change in this area
10:08:54 [Zakim]
10:09:25 [pfps]
pfps: no proposals for extending RDF/XML indicates that there is little need
10:09:39 [Zakim]
10:09:42 [pfps]
guus: subsume changes to RDF/XML under Issue 23
10:09:42 [danbri]
(imho RDF/XML is a pain because RDF is a pain, not because RDF/XML is particularly poorly defined)
10:11:13 [pfps]
mischat: what is the relationship between quad syntaxes and SPARQL's (construct) view of the world
10:11:35 [pfps]
guus: let's put this in as a note on some issue
10:11:39 [mischat]
zakim., who is making noise ?
10:11:41 [ivan]
zakim, who is noisy?
10:11:52 [Zakim]
ivan, listening for 10 seconds I heard sound from the following: Meeting_Room (48%)
10:11:56 [OlivierCorby]
OlivierCorby has joined #rdf-wg
10:11:59 [danbri]
zakim, who else is noisy?
10:11:59 [Zakim]
I don't understand your question, danbri.
10:12:11 [Guus]
10:12:13 [pchampin]
zakim, mute AZ
10:12:13 [Zakim]
AZ should now be muted
10:12:41 [pfps]
cygri: note on issue 30
10:12:58 [mischat]
i will annotate issue-30
10:13:32 [pfps]
guus: we appear to have a reasonable list of issues for graphs
10:13:48 [tomayac]
gavinc: same here :-( back to normal now, though :-)
10:14:00 [pfps]
guus: what should we work on first?
10:15:15 [NickH]
10:15:59 [ww]
re: issue-33 - maybe there is something to be learned from the evopat work out of leipzig. given a graph and a sparql query, produce a sub-graph. that process in some sense identifies the sub-graph.
10:16:02 [pfps]
ivan: what are the notions that we want to standardize?
10:16:28 [pfps]
ivan: let's start with Richard's minimum solution and then critique it
10:18:18 [pfps]
guus: the minimum solution has syntax considerations so let's start there - this is issue 31
10:19:08 [pfps]
guus: Richard had a solution for what to put in to Concepts to handle named graphs
10:19:11 [mischat]
I added a note to re: construct and quads davidwood
10:19:16 [pfps]
cygri: there were comments on that
10:20:03 [pfps]
sandro: my biggest issue is 15 - what is the relationship between IRI and a graph, i.e., what is the basics for semantics of named graphs?
10:20:19 [pfps]
guus is writing down a list of important issues
10:21:50 [ww]
is it a common convention to name graphs with the uri of their "main" subject? doing so helps dereferencing...
10:21:55 [pfps]
issue list - 30: SPARQL dataset; 5: graph literals; 31: syntax; 23: media types; 15: semantics
10:22:37 [pfps]
sandro: we could also try to pick out a small number of motivating use cases
10:22:59 [zwu2]
could not hear anything
10:23:02 [pfps]
guus: do we have all the critical issues
10:23:19 [pfps]
fabian: what about terminology?
10:23:38 [zwu2]
Can somebody please check the phone?
10:23:52 [pfps]
guus: we all agree that the concepts are OK, but the names (g-*) are temporary
10:24:00 [Zakim]
10:24:03 [danbri]
zakim, who is on the phone?
10:24:03 [Zakim]
On the phone I see gavinc, Meeting_Room, OlivierCorby, zwu2 (muted), OlivierCorby.a, OlivierCorby.aa,, OlivierCorby.aaaa, AZ (muted)
10:24:06 [Zakim]
Meeting_Room has David, Wood, Sandro, Mateo, Steve, Harris, Fabien, Pierre, Antoine, Cygri, Yves, Nick, Jean-François, Jan, PFPS, Paul, Groth, Chris, Matheus, Dan, Brickley,
10:24:09 [Zakim]
... Misha, Tuffield, Ivan
10:24:13 [tomayac]
(gavinc and tomayac got disconnected)
10:24:18 [Zakim]
10:24:26 [danbri]
anyone else on the phone that hears us?
10:24:36 [AZ]
I can't hear anything now
10:24:37 [Zakim]
10:24:38 [mischat]
can anyone hear us
10:24:38 [Zakim]
10:24:55 [Zakim]
10:25:03 [tomayac]
(dialed in again, but silence)
10:25:12 [zwu2]
silence for me too
10:25:34 [danbri]
we won't be able to fix it immediately, it seems - sorry
10:25:41 [Zakim]
10:26:09 [pfps]
davidwood: can we close 14 now as being subsumed
10:26:26 [pfps]
guus: let's not do this just now - in any case it may not be completely subsumed
10:26:36 [tomayac]
ivan: thanks. i'll be around till about 13:30 CEST, then need to leave (you know why, feel free to let people know)
10:26:45 [pfps]
cygri: other actions all appear to be less major
10:26:56 [pfps]
guus: action 31 may also be less major
10:27:10 [Zakim]
10:28:19 [pfps]
cygri: issue 28 appear to be subservient to issue 5
10:28:22 [Zakim]
10:28:31 [mischat]
zakim, ??P2 is me
10:28:31 [Zakim]
+mischat; got it
10:28:51 [zwu2]
10:28:57 [pfps]
cycgri: issue 32 appears to be dependent on important one
10:30:04 [pfps]
guus: getting the issues out is an important goal
10:30:32 [pfps]
guus: the breakout should look at at least some of these issues and try to come up with potential solutions
10:30:55 [pfps]
guus: the breakout group should progress on
10:32:32 [danbri]
re graphs, ... Issue 1: If property instances can link only two individuals, how do we deal with cases where we need to describe the instances of relations, such as its certainty, strength, etc? ... do we expect to improve that situation?
10:32:58 [tomayac]
i'll be off for the whole afternoon :-( see you tomorrow. sorry to miss out.
10:33:05 [FabGandon]
FabGandon has left #rdf-wg
10:33:19 [Zakim]
10:33:20 [Zakim]
10:33:48 [Zakim]
10:34:12 [Zakim]
10:43:12 [ww]
bon apetit everyone!
11:32:25 [OlivierCorby]
OlivierCorby has joined #rdf-wg
11:33:41 [cygri]
cygri has joined #rdf-wg
11:36:38 [mbrunati]
mbrunati has joined #rdf-wg
11:37:28 [mischat]
can you guys still here us ?
11:37:41 [mischat]
i think the room is about to dial in now before we start
11:37:41 [sandro]
webcam running again.
11:38:16 [Zakim]
11:38:19 [Zakim]
11:38:27 [Zakim]
11:38:37 [mischat]
gavinc: can you hear properly ?
11:39:06 [pchampin]
sandro: now speaking about deprecation
11:39:36 [pchampin]
in a computer contexte: recommend not to use something that WILL eventually be replaced
11:40:02 [pchampin]
danbri proposed to use the word 'archaic' instead
11:40:08 [mischat]
+1 to dan's language on this topic
11:40:11 [pchampin]
which does not imply replacement
11:40:27 [pchampin]
11:40:46 [danbri]
(I proposed it mainly for vocabulary items; haven't thought about it so much for language-builtin / syntax / grammar aspects)
11:41:19 [pchampin]
ivan: what are the criteria to turn something into archaic' ?
11:41:23 [pchampin]
(and what is the verb meaning 'turning something into archaic' ?)
11:42:05 [pchampin]
sandro : [quoting the proposed text of issue 10]
11:42:26 [pchampin]
ivan: it does not answer my question
11:43:12 [gavinc]
11:43:20 [pchampin]
sandro: who likes proposal on issue-12?
11:43:39 [pchampin]
peter: there might be consequences with the semantics
11:43:53 [sandro]
unanmous support
11:44:01 [pchampin]
11:44:09 [pchampin]
11:45:00 [FabGandon]
FabGandon has joined #rdf-wg
11:46:02 [gavinc]
PROPOSED: Mark xs:string as archaic for use in RDF, recommending use of plain literals instead. Recommend that systems silently convert xs:string data to plain literals.
11:46:03 [pchampin]
sandro: maybe we should not settle on this right now given the short notice
11:46:13 [sandro]
11:46:19 [ivan]
11:46:25 [gavinc]
11:46:25 [mischat]
there was a straw-poll in the room asking if anyone objects to making the xs:string archaic
11:46:26 [pchampin]
guss: we can make a resolution and change it if there are objections
11:46:26 [cygri]
11:46:27 [cmatheus]
11:46:27 [NickH]
11:46:28 [danbri]
11:46:28 [FabGandon]
11:46:28 [mischat]
11:46:29 [pfps]
11:46:29 [davidwood]
11:46:30 [pchampin]
pchampin: +1
11:46:46 [SteveH]
SteveH has joined #rdf-wg
11:46:50 [mischat]
any objections ?
11:47:00 [sandro]
RESOLVED: Mark xs:string as archaic for use in RDF, recommending use of plain literals instead. Recommend that systems silently convert xs:string data to plain literals.
11:47:07 [yvesr]
11:47:37 [davidwood]
Yvesr: It is still minuted :)
11:47:51 [SteveH]
I was +1 too for hte record
11:48:00 [pchampin]
peter: to do it right, we need to check whether it requires some check in the Semantics document
11:48:02 [sandro]
action: peter to make sure the resolution to issue-12 gets into semantics document
11:48:03 [trackbot]
Created ACTION-27 - Make sure the resolution to issue-12 gets into semantics document [on Peter Patel-Schneider - due 2011-04-20].
11:48:30 [sandro]
action: steve make sure issue-12 resolution gets to SPARQL
11:48:30 [trackbot]
Created ACTION-28 - Make sure issue-12 resolution gets to SPARQL [on Steve Harris - due 2011-04-20].
11:48:54 [pchampin]
sandro: now talking about issue-13
11:49:24 [sandro]
subtopic: issue-13
11:49:32 [gavinc]
+q talk about rdf:XMLLiteral support in Jena, Raptor, 4Store, etc
11:49:44 [gavinc]
+q to talk about rdf:XMLLiteral support in Jena, Raptor, 4Store, etc
11:49:45 [pchampin]
ivan: I hate XMLLiterals, but there are valid use cases (e.g. RSS)
11:49:47 [sandro]
11:50:37 [SteveH]
q+ to talk about canonicalisation
11:51:31 [SteveH]
11:51:51 [sandro]
cygri: Maybe just change the canonicalization
11:51:55 [pchampin]
jean-françois: back on issue 12, why not make it the other way? considering plain litteral as a shortcut for xsd:string?
11:52:17 [Zakim]
11:53:05 [mischat]
11:53:06 [pchampin]
sandro: I'm surprised about RSS; I only occasionally looked at RSS, but I saw quoted XML, not XMLLiteral
11:53:20 [davidwood]
ack gavinc
11:53:20 [Zakim]
gavinc, you wanted to talk about rdf:XMLLiteral support in Jena, Raptor, 4Store, etc
11:53:38 [sandro]
sandro: on reason to MAA xmlliteral is that they're often/usually broken
11:53:41 [danbri]
I'm not convinced by the RSS case; RSS1 preceeded xml:Literal by 4 years, and had lost out to RSS2 and Atom by time of RDFCore. Most RSS/Atom feeds are not RDF/XML these days.
11:54:13 [pchampin]
gavinc: are there any implementation that use XMLLiterals properly?
11:54:19 [sandro]
gavin: I'm not convinced XMLLiterals will get any better.
11:55:04 [sandro]
ivan: I'd rather we try to fix rdf:XMLLiteral
11:55:04 [pchampin]
ivan: at the moment, I would prefer to postpone that and see whether the ambiguities of XMLLiteral, relative to canonicalization, can be fixed
11:55:10 [sandro]
11:55:30 [pchampin]
the current canonicalization are not clear on who must do what
11:56:05 [pchampin]
I would be in favor or doing something cleaner IF we can
11:56:34 [pchampin]
sandro: we can do a straw poll about either trying to fix XMLLiteral or dropping them
11:56:43 [gavinc]
in favor of archaic
11:56:49 [sandro]
sense of room --- try to fix it.
11:57:20 [pchampin]
david: why would you mark it as archaic, peter
11:57:29 [sandro]
peter: MAA because it's implemented sooo badly. EG in interacting with OWL.
11:58:14 [pchampin]
peter: requires any OWL parser to have a *working* XML canonicalizer
11:58:30 [pchampin]
subtopic: issue-24 Containers
11:58:35 [gavinc]
also, HTML5 isn't XML ;) so droping it into RDF can't use XMLLiteral
11:58:51 [mischat]
do we want this :
11:59:11 [sandro]
danbri: Bag is useless, Alt is incoherent, Seq doesn't bother me the same way.
11:59:32 [pchampin]
ivan: there is a huge lot of RDF data out there that use containers
11:59:34 [sandro]
ivan: Lots of data using this, but that's okay.
11:59:53 [pchampin]
david: we are not including lists, here
12:00:16 [LeeF]
LeeF has joined #rdf-wg
12:00:17 [sandro]
steve: if rdf Collections were better, I'd be more okay with this.
12:00:21 [sandro]
12:01:16 [pchampin]
jean-françois: part of the problem is that they have no defined semantics
12:01:36 [Zakim]
+ +1.617.553.aagg
12:01:38 [LeeF]
zakim, aagg is me
12:01:38 [Zakim]
+LeeF; got it
12:01:40 [pchampin]
it was planned for the future, but never done
12:02:43 [pchampin]
ivan: the container vocabulary contains all the rdf:_i terms, which are in infinite number
12:02:54 [pchampin]
which causes trouble in the semantics
12:03:21 [LeeF]
RRSAgent, pointer?
12:03:21 [RRSAgent]
12:03:23 [pchampin]
If we MAA (mark as archaic) them, we can simplify the semantics
12:03:25 [ericP]
Zakim, please dial ericP-office
12:03:25 [Zakim]
ok, ericP; the call is being made
12:03:27 [Zakim]
12:03:33 [ericP]
Zakim, please disconnect ericP
12:03:33 [Zakim]
EricP is being disconnected
12:03:34 [Zakim]
12:04:04 [Zakim]
12:04:08 [pchampin]
steve: there are several problems with them
12:04:12 [pchampin]
serializing them in turtle
12:04:19 [pchampin]
no way to close them
12:04:47 [pchampin]
guus: is there a way to fix some of them?
12:05:08 [danbri]
soemthing like: "The originally specified meanings of rdf:Alt and rdf:Bag constructs have not proved generally useful; rdf:Seq has more utility, but shares some formal problems with the others. They are all considered archaic constructs."
12:05:25 [pchampin]
sandro: on the other hand, they are handy with SPARQL
12:05:28 [mischat]
does anyone want to keep rdf alt and rdf bag ?
12:05:38 [sandro]
every want to bag Alt and Bag.
12:06:26 [danbri]
steveh, that list-as-datatype ... written up somewhere?
12:06:41 [SteveH]
danbri, no
12:06:42 [pchampin]
ivan: we have to be careful vis a vis Adobe how we mention that alt is now archaic
12:07:24 [pchampin]
XMP uses Alt and Seq
12:07:56 [danbri]
q+ to suggest an action on ivan to blog this
12:08:10 [danbri]
12:08:49 [pchampin]
sandro: if we found a better way to do it, would you be ok to get rid of Seq?
12:09:16 [sandro]
general sense that we should MAA rdf:Seq *if* we have a sensible alternative.
12:09:23 [pchampin]
[a majority of hands raised]
12:10:10 [danbri]
12:10:13 [sandro]
zakim, who is on the call?
12:10:13 [Zakim]
On the phone I see OlivierCorby, OlivierCorby.a, OlivierCorby.aa,, gavinc, OlivierCorby.aaaa, Meeting_Room, zwu2, LeeF, EricP
12:10:25 [pchampin]
topic: Turtle TF
12:11:37 [pchampin]
steve: Level of SPARQL compatibility (issue-1)
12:11:52 [pchampin]
keywords (prefix, base)
12:11:55 [pchampin]
number handling
12:12:30 [pchampin]
(issue-18 what does "18." mean)
12:12:47 [pchampin]
yves: lots of parsers will return different things in Turtle
12:13:11 [pchampin]
steve: the SPARQL solution is that "18." is a decimal
12:13:33 [pchampin]
you need a space to put a dot after 18 as an int
12:14:04 [pchampin]
sandro: I feel that the space before the dot is making it hard for people to adopt Turtle
12:14:20 [pchampin]
ivan: it makes it hard for me :) I always forget it
12:14:55 [pchampin]
sandro: why not require a zero after the dot if you want a decimal?
12:15:03 [pchampin]
steve: having SPARQL change that is not an issue
12:15:30 [pchampin]
david: it's a purely syntactical point that some people feel religious about
12:15:35 [pchampin]
possibly for no good reason
12:16:00 [gavinc]
Why the heck is 18. a decimal in the first place?
12:16:16 [mischat]
because of the xml spec iirc gavinc
12:16:28 [gavinc]
12:16:33 [ericP]
i think there's a lot of precedent for that in existing programming langs
12:16:48 [mischat]
12:16:51 [pchampin]
jan: this is linked to another problem: local name ending with a dot
12:17:02 [pchampin]
steve: this is in a further slide
12:17:10 [sandro]
cygri: Can't you tell from the grammar?
12:17:26 [pchampin]
steve: it makes the grammar more compocated to implement
12:17:31 [pchampin]
12:17:32 [sandro]
SteveH: It's hard, it might be like lookahead 2 or something.
12:17:50 [sandro]
+1 steve: require the zero, and the SPARQL folks to fix it, it was a bug.
12:18:07 [gavinc]
I don't think 18. is vaild in XQuery... if it is... I sure as heck never saw it
12:18:21 [sandro]
(and note that SPARQL folks can stull use the .0 )
12:18:32 [pchampin]
12:18:46 [ericP]
+1 to steve's "just fix it in turtle" proposal
12:19:35 [gavinc]
+1 to just fix it in turtle
12:19:41 [mischat]
it is being proposed that removing the trailing "." at the end of turtle statements (via jaan) would be an easier fix
12:20:10 [ivan]
12:20:37 [sandro]
PROPOSED: close ISSUE-18 by requiring digits after the decimal point, as in "18.0".
12:20:38 [sandro]
12:20:48 [mischat]
12:20:48 [sandro]
steve: +1
12:20:49 [ivan]
12:20:51 [gavinc]
12:20:53 [mischat]
+1 from steveH
12:20:57 [davidwood]
12:20:57 [NickH]
12:20:59 [zwu2]
12:21:02 [pchampin]
pchampin: +1
12:21:09 [FabGandon]
12:21:12 [cygri]
12:21:28 [gavinc]
12:21:34 [gavinc]
There is a whitespace in turtle issue ;)
12:21:44 [ericP]
12:21:45 [JFB]
JFB has joined #rdf-wg
12:21:51 [sandro]
RESOLVED: close ISSUE-18 by requiring digits after the decimal point, as in "18.0"
12:22:07 [LeeF]
12:22:08 [cmatheus]
12:22:14 [danbri]
(aside: I was just thinking: lifetime of average Turtle document is likely somewhat longer than lifetime of average SPARQL query)
12:22:24 [sandro]
ACTION: Lee to convey our resoltuon on ISSUE-18 to SPARQL WG
12:22:24 [trackbot]
Created ACTION-29 - Convey our resoltuon on ISSUE-18 to SPARQL WG [on Lee Feigenbaum - due 2011-04-20].
12:22:48 [pchampin]
steve: issue-1 qnames
12:23:06 [pchampin]
legal in SPARQL: ns:123 ns:1.2 ns:aaa.bbb
12:23:11 [davidwood]
Danbri: Perhaps not with SPARQL stored procedures.
12:23:12 [pchampin]
not legal: ns:aaa.
12:23:40 [mischat]
this was motivated due to dots in filenames
12:23:52 [pchampin]
not sure about what turtle exactly saus
12:23:56 [pchampin]
12:24:23 [ericP]
12:24:40 [ericP]
12:24:49 [ericP]
( PN_CHARS_U | [0-9] ) ( ( PN_CHARS | "." )* PN_CHARS )?
12:25:29 [gavinc]
doesn't sparql/turtle also allow digits where NCName doesn't?
12:25:48 [pchampin]
peter: [detailed account of the differences btw SPARQL and Turtle re qnames]
12:26:19 [ericP]
gavinc, yes it does. NCName prohibits leading digits in the localname
12:26:58 [pchampin]
david: remark that neither SPARQL nor Turtle refere to the definition of QNames
12:27:12 [pchampin]
which is restricted by the XML syntax
12:27:52 [pchampin]
guus: who objects to copying the SPARQL definition into turtle?
12:28:12 [pchampin]
ivan: bringing them as close as possible is a good thing
12:28:24 [zwu2]
keep things consistent is good
12:28:32 [LeeF]
We got very strong comments from life sciences folks before we made this change in SPARQL 1.0
12:28:37 [pchampin]
sandro: I'm not fond of SPARQL identifiers, which are too persmissive re programming language identifiers
12:28:39 [danbri]
q+ to ask i18n/l18n concerns
12:28:43 [LeeF]
I can find those comments if that would be useful to anyone
12:28:51 [gavinc]
so does RDF/XML
12:28:53 [pchampin]
12:28:59 [gavinc]
Javascript doesn't
12:29:21 [Guus]
12:29:27 [pchampin]
peter: anyway, SPARQL and Turtle accept unicode characters, so most programming languages are waaay behind already
12:29:53 [pchampin]
@prefix ☃: <>
12:29:56 [danbri]
12:29:57 [LeeF]
ACTION-29: See
12:29:58 [trackbot]
ACTION-29 Convey our resoltuon on ISSUE-18 to SPARQL WG notes added
12:30:14 [danbri]
12:30:26 [danbri]
vs kowiki:암스테르담
12:30:33 [gavinc]
heh, yeah, don't do that.
12:31:01 [pchampin]
cygri: I would not be surprised that programming language have different restrictions, anyway
12:31:01 [gavinc]
having just written a python library it ended badly. Way better off using object['blah'] notation for RDF
12:31:38 [pchampin]
danbri: aren't we hindering i18n here?
12:31:39 [danbri]
12:31:47 [pchampin]
ivan: no, we are extending the space of legal things
12:32:01 [davidwood]
davidwood has joined #rdf-wg
12:32:28 [davidwood]
LeeF: right. Thanks.
12:32:31 [sandro]
PROPOSED: Allow dots inside local part of qnames in Turtle, aligning with SPARQL syntax
12:33:32 [davidwood]
12:33:49 [davidwood]
(from SPARQL)
12:33:50 [sandro]
+0 (I like having qnames line up with legal field names in programming languages)
12:34:01 [LeeF]
12:34:03 [cygri]
+1 to do what sparql does
12:34:04 [ericP]
+1 (i've given up)
12:34:05 [cmatheus]
12:34:07 [sandro]
PROPOSED: Allow dots inside local part and namespace part of qnames in Turtle, aligning with SPARQL syntax
12:34:07 [zwu2]
12:34:08 [gavinc]
12:34:16 [pfps]
12:34:18 [mbrunati]
12:34:19 [ivan]
12:34:19 [pchampin]
pchampin: +1
12:34:20 [NickH]
0 (as a ruby user)
12:34:24 [sandro]
+0 (I like having qnames line up with legal field names in programming languages)
12:34:29 [FabGandon]
12:35:03 [davidwood]
12:35:07 [yvesr]
12:35:16 [sandro]
RESOLVED: Allow dots inside local part and namespace part of qnames in Turtle, aligning with SPARQL syntax
12:35:31 [danbri]
(so kowiki:암스테르담 is ok?)
12:35:41 [gavinc]
12:35:41 [ericP]
12:36:15 [pchampin]
steve: continuing on issue-1: features
12:36:19 [ww]
+0.5 belatedly
12:36:21 [mischat]
who wants to add more features to turtle ?
12:36:32 [pchampin]
stick to the feature-set in submission?
12:36:47 [LeeF]
I think these ought to be different discussions.
12:36:50 [gavinc]
-1 to adding features +0 to TriG as part of Turtle
12:36:54 [pchampin]
or add more: quads? inverse paths? equals? more sugar?
12:37:03 [LeeF]
Discussion #1: Is Turtle extended to handle named graphs?
12:37:07 [mischat] for visual purposes
12:37:15 [LeeF]
Discussion #2: Does Turtle have other features from N3, elsewhere?
12:37:47 [pchampin]
yves: does this have implications like property paths?
12:37:53 [davidwood]
gavinc: "-1" is a formal objection. Is that your intent?
12:37:57 [LeeF]
mischat, that diagram is great, thanks.
12:37:58 [gavinc]
12:38:08 [gavinc]
12:38:09 [pchampin]
ivan: I'm scared by these questions
12:38:11 [LeeF]
mischat, It's missing "is ... of ... ", right?
12:38:11 [gavinc]
12:38:25 [Guus]
12:38:26 [sandro]
yeah, gavinc
12:38:27 [pchampin]
would require a lot of rewriting in deployed parsers
12:38:27 [NickH]
has anyone implemented any extra features in their turtle parser?
12:38:38 [yvesr]
i was thinking of :a foaf:knows/foaf:lnows :b <=> :a foaf:knows _:c . _:c foaf:knows :b
12:38:38 [ww]
i would like to see in turtle, = shorthand for owl:sameAs and trig means <graph> = { ... } where the = can be omitted for brevity
12:39:04 [ww]
this preserves compatibility with trig and gives a path towards n3
12:39:06 [sandro]
zakim, who is on the call?
12:39:06 [Zakim]
On the phone I see OlivierCorby, OlivierCorby.a, OlivierCorby.aa,, gavinc, OlivierCorby.aaaa, Meeting_Room, zwu2, LeeF, EricP
12:39:14 [mischat]
12:39:32 [pchampin]
guus: we are likely to include a separate media type that extend turtle with named graphs
12:39:41 [davidwood]
ack Guus
12:39:46 [pchampin]
what if we include any new feature in this new media type?
12:39:53 [danbri]
q+ to suggest we structure the HTML of the document to encourage re-use of productions from the turtle grammar
12:40:16 [pchampin]
cygri: if we do that, this will drive most people to the multi-graph format just for the benefit of the other features
12:40:24 [pchampin]
pchampin: +1 cygri
12:41:00 [sandro]
steve: no requests for this stuff from 4store users
12:42:05 [pchampin]
mischa: people will have to rewrite turtle parsers anyway, re changes in prefix
12:42:22 [pchampin]
ivan: but inverse paths are a much deeper change in the parser
12:43:32 [pchampin]
danbri: sugar for inverse path is also in RDFa
12:43:35 [davidwood]
12:43:43 [danbri]
ack q?
12:43:50 [danbri]
ack danbri
12:43:50 [Zakim]
danbri, you wanted to suggest we structure the HTML of the document to encourage re-use of productions from the turtle grammar
12:43:53 [davidwood]
ack danbri
12:44:37 [pchampin]
danbri: basically SPARQL and Turtle are the same
12:44:50 [sandro]
danbri: I'd like to see these features, but I don't think they need to be this Turtle spec.
12:45:59 [pchampin]
guus: points out that the quad extension is a separate issue
12:46:36 [pchampin]
it is a shame Nathan is not here to discuss the matter
12:46:48 [pchampin]
we can phrase a resolution and put it on the agenda of the next telecon
12:47:28 [mischat]
FabGandon: sorry, i don't know why that got in the diagram, and I don't really parse the "x!y^z paths", and i don't know much about n3 either
12:47:33 [sandro]
PROPOSED: Our turtle will have the same feature-set as the submission (leaving out inverse paths, leaving out "=", and other N3 things)
12:48:59 [sandro]
ISSUE: Do we need to add features to turtle, beyond what's in the Submission (such as inverse paths and =)?
12:49:06 [trackbot]
Created ISSUE-34 - Do we need to add features to turtle, beyond what's in the Submission (such as inverse paths and =)? ; please complete additional details at .
12:49:33 [sandro]
Guus: This is NOT pre-judge solution to GRAPHs.
12:49:52 [cygri]
LeeF, I count 11 macs and 6 others in the room. scary!
12:51:04 [sandro]
action: guus put issue-34 on agenda for next time, proposed resolution "No"
12:51:04 [trackbot]
Created ACTION-30 - Put issue-34 on agenda for next time, proposed resolution "No" [on Guus Schreiber - due 2011-04-20].
12:51:37 [pchampin]
steve: syntaxes (issue4, 31etc)
12:51:54 [mischat]
sandro: perhaps moving the webcam to face the screen
12:52:08 [pchampin]
triples+terse = turtle
12:52:19 [pchampin]
triples+verbose = NTriples
12:52:34 [pchampin]
quads+terse = trig,n3,sparql update,qurtle
12:52:36 [LeeF]
triples+verbose = vertle, naturally
12:52:42 [pchampin]
quad+verbose: NQuads
12:53:52 [pchampin]
ivan: there is a non-trivial difference btw turtle and ntriples: the latter only accept ascii
12:54:00 [danbri]
grep '^quad' /usr/share/dict/words >>>
12:54:41 [pchampin]
guus: should we put in our spec what are the restrictions on NTriples
12:54:48 [pchampin]
as a section in the Turtle document
12:55:08 [pchampin]
steve: with a more rational media type than text/plain
12:55:46 [pchampin]
cygri: I would argue to have a separate document, as they describe rather different formats
12:55:46 [mischat]
q+ about bnode serialisation and ordering of documents
12:55:53 [pchampin]
peter: I would argue against that
12:56:13 [mischat]
12:56:47 [pchampin]
we would have too many documents
12:57:01 [pchampin]
cygri: with RDFa and JSON, we will have multiple documents anyway
12:57:17 [pchampin]
ivan: what problem are we trying to solve here?
12:57:37 [pchampin]
ntriples has been around for some time? in a W3C recommendation?
12:57:43 [pchampin]
what do we need to fix?
12:58:02 [sandro]
cygri: if I google for N-Triples, I end up in the wrong place.
12:58:15 [danbri] is 1st hit, and it says 'PLEASE NOTE: This document has been superceded by the RDF Test Cases Working Draft. See N-Triples for more information.'
12:58:50 [zwu2]
12:59:13 [pchampin]
guus: for the purpose of editing the recommendation, it makes more sense to have ntriples as an appendix of turtle
12:59:34 [pchampin]
we can also have an ntriple primer pointing to that appendix
12:59:55 [pchampin]
ivan: from what Richard said, this is just an editorial issue, so postpone
13:00:31 [pchampin]
danbri: who greps ntriple on a daily basis
13:00:40 [pchampin]
quite a few hands raise
13:01:12 [pchampin]
steve; and gets bitten by the fact that it is suppose to be ascii, and is often utf8 in practice
13:02:02 [danbri]
aside re naming -- I've googled all the words that begin ^quad; nothing great. Trying with ^trip -> is interesting (lots of meanings but none clash)
13:03:34 [ericP]
+1 to deprecating ntriples
13:03:57 [pchampin]
paul: if we redefine ntriples as a subset of turtle, don't all those issues disappear?
13:04:39 [davidwood]
ack mischat
13:04:41 [pchampin]
steve: yes, mostly
13:04:42 [mischat]
13:04:47 [mischat]
cygri: ^^ ?
13:04:52 [cygri]
13:04:59 [cygri]
oh thanks mischat
13:05:02 [zwu2]
-1 to changes to ntriples
13:05:16 [cygri]
zwu2: including not defining a media type for it?
13:05:30 [zwu2]
13:06:09 [pchampin]
zwu2: we will officially object any change to ntriples
13:06:18 [danbri]
maybe the issue here is 'change'
13:06:41 [pchampin]
david: are you parsing ntriples in ascii? are you sure?
13:07:16 [mischat]
on the wiki page above ^^
13:07:28 [zwu2]
ack zwu2
13:07:33 [pchampin]
danbri thinks ntriples-is-ntriples; whatever this group does is ... the next thing
13:07:37 [davidwood]
zwu2 confirmed that they would formally object to a change in ntriples
13:07:39 [ivan]
13:07:41 [zwu2]
13:07:43 [sandro]
maybe leave N-Triples alone and define "Line-Mode Turtle" as the relevant Turtle subset?
13:08:01 [davidwood]
Guus would prefer not to do that...
13:08:09 [pchampin]
steve: about the second line (quads + terse/verbose)
13:08:24 [gavinc]
n-quads too!
13:08:37 [pchampin]
there are a lot of turtle-like languages for quads
13:09:01 [pchampin]
ivan: sparql update?
13:09:26 [pchampin]
steve: yes, sparql update allows you to express graphs that ends up being stored, so it is a serialization syntax of its own
13:10:44 [Zakim]
13:10:45 [pchampin]
guus: iirc, richard argued agains having qurtle and turtle defined in the same document
13:11:32 [pchampin]
cygri: as discussed this morning, if turtle is extended with quads, this will have major impact on implementation
13:11:49 [pchampin]
so qurtle (or anything) needs a separate media type and a separate document
13:11:52 [LeeF]
+1 to keeping graph serialization separate from turtle
13:13:01 [pchampin]
peter: I would prefer people consuming turtle to be ready to consume quads, though I don't expect agreement on that
13:13:10 [davidwood]
13:14:09 [pchampin]
paul: we are moving from a specification with triples to a specification with quads
13:14:22 [gavinc]
-1 to a turtle media type document containing more then one graph +0 to the ONLY difference being the media type
13:14:26 [pchampin]
so why not including quads in next-turtle
13:14:27 [webr3]
webr3 has joined #rdf-wg
13:15:35 [gavinc]
Not very worried about HTTP GET, a bit more worried about HTTP POST/PUT
13:16:30 [pchampin]
steve: what if you crawl untrusted documents, and they contain named graphs?
13:17:03 [pchampin]
naming the graph with a URI that you care about
13:17:53 [pchampin]
danbri: I can answer from an experience
13:17:54 [Guus]
Guus has joined #rdf-wg
13:18:09 [pchampin]
I took some examples about provenance
13:18:26 [pchampin]
converted it to quads
13:18:28 [gavinc] if the representation contains more then one graph, exactly what to do with these updates becomes very strange.
13:19:33 [webr3]
if you put quads or ng's on the web for follow your nose, then I need a 5-tuple store (then if you put that online, will need 6-tuples, etc)
13:19:43 [pchampin]
then was quite confused about the way to consume them,
13:20:18 [mischat]
<snap> gavinc
13:20:22 [danbri]
I made some tests with rdfa+@graph, see"> ... it's pretty confusing to get a sane processing model
13:20:26 [mischat]
13:20:27 [pchampin]
as if the blog says 'this comes from the NYT', I don't want to credit the NYT with it
13:20:27 [mischat]
13:20:36 [danbri]
( parser output )
13:22:17 [pchampin]
steve: the rest of the slides is about sub-details of all the above
13:23:08 [pchampin]
Note that NQuads is not strictly a subset of anything else.
13:23:15 [LeeF]
N-Quads can't serialize empty graphs.
13:23:57 [LeeF]
Or if it can, I'd like to see how?
13:24:42 [pchampin]
peter: you can state that <uri> a :EmptyGraph
13:24:45 [webr3]
empty graph is just <x> a Graph . surely, you know you have a graph, and a name for it, but nothing else
13:25:20 [pchampin]
ivan: back to NQuads, there is a broken symetry here
13:25:41 [sandro]
not at all web3r. Knowing a graph is empty is quite different from not knowing whether it is empty.
13:26:03 [webr3]
good point
13:26:11 [gavinc]
Open World ;) No it isn't
13:26:19 [pchampin]
cygri: Trig and NQuads basically reuse a big part of Turtle (terms) and adds a few production rules around them.
13:26:24 [webr3]
double good point lol
13:27:27 [pchampin]
cygri: I don't think it is essential to many people that NTriples is a subset of turtle; same for NQuads
13:28:17 [gavinc]
Binary RDF!! Bring it on!
13:28:18 [pchampin]
paul: the absence of symetry makes it harder to teach
13:28:32 [LeeF]
Is N-Quads in current use?
13:28:35 [pchampin]
cygri: we have a lot of things on the table that make it even harder to teach
13:28:41 [LeeF]
Or do we have flexibility to (re-)define it?
13:28:55 [zwu2]
yes, Oracle is using n-quads
13:28:59 [LeeF]
zwu2, thanks
13:29:05 [gavinc]
Yes, TopQuadrant is using N-Quads
13:30:09 [pchampin]
mischat: NQuads is very easy to parse and generate,
13:30:30 [pchampin]
while most Trig parsers I have tried do not work well with big files
13:30:46 [gavinc]
Where "big" is tiny
13:30:59 [zwu2]
+1 to mischat
13:31:14 [LeeF]
We work pretty regularly with large TriG files, without much difficulty.
13:31:55 [gavinc]
Lee, 60 million+ triples?
13:32:17 [gavinc]
well, quads ;)
13:32:24 [LeeF]
I think it's a mistake to just do N-Quads. There is real value to human-convenient syntax. We've seen that over and over with turtle (vis a vis N-triples). I don't think it's any different for quads.
13:32:53 [LeeF]
gavinc, yes, I believe so, though I can ask around for particular details
13:32:54 [pchampin]
cygri: the SPARQL document manages to describe datasets without a standard syntax
13:34:29 [pchampin]
guus: by avoiding the quads+terse box, we lose symetry, but we normalize what is already out there
13:35:34 [cygri]
A tree with Turtle as the root, and three children "TriG/SPARQL Update", "N-Triples", "N-Quads"
13:36:07 [davidwood]
13:37:10 [pchampin]
steve: SPARQL update is very similar to TriG, except it has keyword GRAPH in front of the graph URI
13:38:57 [mischat]
mailing list
13:39:13 [webr3]
can you do anything with one that you cannot do with the other? (re trig/n-quads)
13:39:30 [MacTed]
MacTed has joined #rdf-wg
13:39:35 [gavinc]
Yes, a human can read and write TriG ;)
13:39:42 [SteveH]
"Turtle should remain as a syntax only for Triples, some other syntax should be defined to represent quad data"
13:39:45 [LeeF]
13:40:17 [gavinc]
You can also do horrible things with awk and sort to N-Quads ;)
13:40:22 [yvesr]
13:41:02 [mischat]
13:41:04 [pchampin]
PROPOSED: Turtle should remain as a syntax only for Triples, some other syntax should be defined to represent quad data
13:41:09 [webr3]
13:41:11 [cygri]
13:41:11 [yvesr]
13:41:11 [LeeF]
13:41:11 [gavinc]
13:41:15 [zwu2]
13:41:16 [SteveH]
13:41:16 [ivan]
13:41:17 [mischat]
13:41:17 [mbrunati]
13:41:19 [sandro]
13:41:19 [mischat]
13:41:20 [ericP]
13:41:21 [pfps]
13:41:23 [FabGandon]
13:41:23 [NickH]
13:41:24 [pchampin]
pchampin: +1
13:41:24 [davidwood]
13:41:24 [cmatheus]
13:41:27 [danbri]
13:41:36 [sandro]
really -0
13:41:47 [sandro]
(I prefer one syntax with graph literals or something)
13:41:47 [JFB]
13:41:56 [ericP]
⧻1 ?
13:42:06 [webr3]
sandro, +1, n3 like though, not quad like for me
13:42:24 [gavinc]
13:42:29 [sandro]
<> foaf:hates sandro:ericP.
13:43:14 [zwu2]
nice conference room
13:43:22 [pchampin]
ivan: before we take a break and go to JSON,
13:43:36 [pchampin]
I would like to talk about the documentation style of Turtle
13:43:56 [pchampin]
about which Peter and Erik disagreed longly
13:44:24 [mischat]
13:44:26 [davidwood]
13:44:26 [ericP]
i'm sympathetic to pfps's debugging point
13:44:29 [pchampin]
guus: as it is an editorial problem, I think we can postpone it
13:44:31 [gavinc]
15 minutes?
13:44:38 [cygri]
13:44:45 [mischat]
manu1: json stuff when we get back
13:44:47 [mischat]
in 20 mins
13:45:29 [Zakim]
13:46:09 [manu]
zakim, code?
13:46:09 [Zakim]
the conference code is 733941 (tel:+1.617.761.6200 tel:+ tel:+44.203.318.0479), manu
13:46:11 [Zakim]
13:46:11 [Zakim]
13:46:23 [Zakim]
13:46:30 [Zakim]
13:46:31 [Zakim]
13:46:38 [manu]
zakim, I am ??P1
13:46:38 [Zakim]
+manu; got it
13:53:50 [NickH]
14:03:21 [Zakim]
14:03:26 [webr3]
zakim, i am IPcaller
14:03:26 [Zakim]
ok, webr3, I now associate you with [IPcaller]
14:04:34 [Zakim]
+ +1.603.897.aahh
14:05:05 [Zakim]
14:05:19 [Zakim]
14:05:23 [gavinc]
Yes, yes they do. Have we ALL used that for Triple stores now? :D
14:05:51 [webr3]
those ec2 gpu powered instances are awesome
14:06:07 [gavinc]
Intels new cpu supports 256 GB of ram :D
14:06:15 [webr3]
manu, outpace this:
14:07:29 [pfps]
topic: JSON task force
14:07:39 [Zakim]
14:08:28 [sandro]
scribe: NickH
14:08:44 [NickH]
matteo: has been tracking the disussions using a mind map
14:08:49 [mischat]
do we have a link to the slides ?
14:09:13 [NickH]
slides are not currently on the web
14:09:49 [sandro]
zakim, list attendees
14:09:49 [Zakim]
As of this point the attendees have been +1.707.861.aaaa, gavinc, +1.404.978.aabb, tomayac, +31.20.592.aacc, +, OlivierCorby, AZ, David, Wood, Sandro, Mateo, Steve,
14:09:51 [NickH]
matteo is mailing the slides now
14:09:53 [Zakim]
... Harris, Fabien, Pierre, Antoine, Cygri, Yves, Nick, Jean-François, Jan, PFPS, Paul, Groth, Chris, Matheus, Dan, Brickley, Misha, Tuffield, Ivan, +1.408.642.aaff, zwu2,
14:09:55 [sandro]
zakim, who is on the call?
14:09:55 [Zakim]
... mischat, Meeting_Room, +1.617.553.aagg, LeeF, EricP, manu, [IPcaller], +1.603.897.aahh, PatH
14:09:59 [Zakim]
On the phone I see manu, [IPcaller], +1.603.897.aahh, zwu2, PatH, LeeF, OlivierCorby, OlivierCorby.a, OlivierCorby.aa,, gavinc, Meeting_Room
14:10:07 [Souri]
Souri has joined #rdf-wg
14:10:37 [ivan]
ivan has joined #rdf-wg
14:12:02 [NickH]
matteo: there are two presentations, the second presentation is Thomas's one
14:12:42 [Guus]
Guus has joined #rdf-wg
14:13:04 [NickH]
matteo: has made a timeline from the start of the discussions (slide 4)
14:13:08 [zwu2]
Did matteo send out slides to the wg mailing list?
14:14:09 [cygri]
slides attached here:
14:14:23 [NickH]
matteo: on the 6th of march manu produced the JSON design requirements
14:14:24 [zwu2]
14:14:41 [NickH]
matteo: there were two main reactions
14:15:03 [mischat]
14:16:01 [NickH]
1. make a simple way to transform JSON objects into RDF
14:16:07 [Zakim]
14:16:19 [NickH]
2. to provide an RDF serialisation in JSON
14:16:23 [AZ]
Zakim, Ronald is me
14:16:23 [Zakim]
+AZ; got it
14:16:29 [manu]
zakim, mute Ronald
14:16:29 [Zakim]
sorry, manu, I do not know which phone connection belongs to Ronald
14:16:29 [AZ]
zakim, mute me
14:16:30 [Zakim]
AZ should now be muted
14:16:36 [cygri]
zakim, mute them all
14:16:36 [Zakim]
I don't understand 'mute them all', cygri
14:17:17 [mischat]
14:17:22 [mischat]
json market segments ^^
14:17:28 [mischat]
14:18:04 [ivan]
ivan has joined #rdf-wg
14:18:31 [NickH]
matteo: there was a Seperate Call for the JSON Taskforce
14:18:56 [NickH]
matteo: and separented the examples into two main groups
14:19:03 [NickH]
1. Government/Enterprice
14:19:15 [NickH]
2. Independent Web Developer
14:19:42 [mischat] <-- elephant in the room thread
14:19:51 [NickH]
matteo: At the end of the March there was an interesting discussion on the mailing list about 'What *is* JSON'
14:20:27 [PatH]
PatH has joined #rdf-wg
14:21:40 [cygri]
Souri, the webcam can't zoom ... the slides are here, in an attachment:
14:23:35 [NickH]
matteo: (slide 5) looking at the existing work on JSON
14:23:40 [NickH]
no single input document
14:24:08 [NickH]
(slide 6) looking at the use cases for JSON + RDF
14:26:08 [NickH]
matteo: the use cases have made it clearer what the job of the Task Force is
14:27:28 [NickH]
(slide 7) there are two open issues in the tracker about what the starting point and source of JSON specification reference
14:29:54 [mischat]
fwiw, i generated a list of triplestores and the RDF serialisations they support, this includes current practice in the world of JSON RDF
14:29:56 [mischat]
14:30:00 [ivan]
Slides on the web (and not as an attachment):
14:30:03 [NickH]
matteo: tomorrow I am going to try and complete the mind-map that will issulrate all the different issues and serialisations approaches
14:30:27 [ivan]
14:31:08 [NickH]
guus: I am not sure what issues are open for discussion now
14:32:03 [manu]
I really like Sandro's simplification on the big Level/Group thing
14:32:32 [Souri]
14:32:53 [NickH]
looking at:
14:32:53 [NickH]
14:33:10 [NickH]
sandro: Matrix is too complex
14:33:18 [NickH]
sandro: lets focus on what the users want
14:34:14 [NickH]
sandro: there is also a group D - users who want some of RDF, but not all RDF features (eg blank nodes)
14:34:28 [webr3]
subset-of-B appears to be A ..
14:34:29 [manu]
14:34:38 [NickH]
sandro: Group A and C wants something that looks like JSON
14:34:47 [ivan]
ack manu
14:34:48 [NickH]
sandro: Group B and D want something that looks like RDF
14:35:37 [NickH]
manu: I believe we can do Group A,B,C,D as a single serialisation
14:35:53 [manu]
Here's how we can do Group B: [{}, {}, {}]
14:36:01 [manu]
Group A: {}
14:36:25 [NickH]
sandro: I am giving up on trying to solve for all groups with a single solution
14:36:52 [gavinc]
A) Non RDF aware developers B) RDF aware developers?
14:37:00 [NickH]
matteo is now continuing with his slides
14:37:44 [zwu2]
sandro, does group C has a JSON view like B?
14:38:55 [sandro]
zwu2, Group C uses an API to get RDF-triples. It doesn't really care what the JSON looks like.
14:39:29 [webr3]
minor point: everybody who follows their nose around the web of data requires an API regardless, they have to
14:39:30 [NickH]
(slide 12) shows a document from several years ago that compares and contrasts XML and RDF
14:39:30 [pgroth]
just curious - does anyone have a good pointer to a "popular" site that pushes json based on an rdf backend?
14:39:44 [zwu2]
I see, Sandro
14:39:59 [PatH]
Slide 12 is cute
14:40:04 [sandro]
pgroth, maybe some of the UK Gov't stuff using Linked Data API.
14:40:10 [NickH]
matteo: perhaps we should make a similar diagram explaining the differences between RDF and JSON
14:40:27 [cygri]
pgroth, dbpedia?
14:40:46 [gavinc]
not popular, but
14:40:54 [pgroth]
cygri, do lots of people use the json end of dbpedia?
14:41:06 [NickH]
pgroth: the BBC World Cup website was built using JSON from a Triplestore
14:41:19 [cygri]
pgroth, i have no numbers about that
14:41:33 [webr3]
pgroth, nytimes
14:41:59 [pgroth]
webr3, best example so far :-)
14:42:34 [gavinc]
Huh, I had no idea the nytimes RDF came as JSON too
14:42:55 [mischat]
14:43:12 [pgroth]
that is some ugly urls
14:43:20 [NickH]
now showing "JSON Syntax Options" PDF by Thomas Steiner
14:43:27 [yvesr]
webr3: i wonder why that example doesn't use native json datatypes (for lat/long)
14:43:30 [mischat]
is the syntax a named syntax, the nytimes one ? does it related to any of the json syntaxes up on our wiki?
14:43:52 [SteveH]
jsonlint seems to believe that the / escaping is neccesary
14:44:04 [webr3]
yvesr, I'm unsure, it doesn't use /any/ native types
14:44:35 [NickH]
14:45:11 [danbri]
ivan, " Allowing \/ helps when embedding JSON in a <script> tag,"
14:46:45 [webr3]
comments re nytimes data: note no nested objects, subject as key, all values always in an array (this is something Joe Presebrey from MIT also said was most useful for consuming), doesn't use any native types, bad slashing style, uses namespaces
14:46:52 [zwu2]
like the matrix one
14:48:13 [NickH]
first, a object based approach, which is close to what Web Developers expect from JSON
14:48:31 [mischat]
there is a widely deployed json-triple format in the wild at the moment iirc
14:48:34 [NickH]
second, a triple-based approach which is very close to RDF
14:48:47 [webr3]
more comments re nytimes: this actually appears to be more machine optimized than developer friendly (looses dot notation, no usage of basic types, will require pre-processing before using in most cases)
14:50:01 [NickH]
sandro: is confused by the _:id in the example
14:50:06 [manu]
Sandro: We need to discuss why we used '@' - there was a reason :P
14:50:23 [NickH]
not clear if id is a real string or a placeholder
14:50:24 [webr3]
re "Findings - subjects (object-based)" the first allows nested objects, the second pretty much precludes it
14:50:50 [danbri]
IRIs can contain spaces, right?
14:51:03 [NickH]
sandro: it is totally necessary to be able to distinguish between URIs and strings
14:51:04 [PatH]
Is the example being discussed web-visible?
14:51:16 [webr3]
14:51:25 [NickH]
sandro: don't take the lack of mails on the mailing list as agreement
14:51:26 [webr3]
(slide 11 now)
14:51:31 [PatH]
14:52:27 [manu]
It says "Findings - IRIs" at the top, NickH
14:52:43 [NickH]
short dicussion about wether IRIs can contain whitespace
14:52:49 [manu]
Also - I have a solution for the language tags and IRIs via @context...
14:53:05 [manu]
haven't had a chance to put that out to the mailing list - we could have a trigger in the @context
14:53:11 [NickH]
webr3: it isn't on the projection
14:53:25 [manu]
"@context" : { "@microsyntaxes" : true }
14:53:26 [sandro]
(I really really dont like the microsyntax approaches.)
14:53:43 [NickH]
question about how to escape @ sign in the micro format for languages
14:53:51 [danbri]
14:53:52 [manu]
I don't like the Microsyntax approaches either, but when you take the Web-developer-friendly object-based approach, you paint yourself into a corner.
14:54:22 [NickH]
SteveH: why is '@' being used for datatypes as well as languages
14:54:30 [manu]
14:54:39 [webr3]
unsure whether lang or datatype necessary for /all/ use cases..
14:54:52 [NickH]
sandro: doesn't like the Microformat approach, don't like parsing both
14:54:57 [ivan]
ack manu
14:55:20 [sandro]
sandro: I think microparsing is the worst of both worlds....
14:55:25 [NickH]
manu: agrees but thinks microformats may be needed for the edge cases
14:55:31 [cygri]
sandro +1
14:56:13 [PatH]
meta-comment, we ought to avoid trying to police how the planet uses JSON...
14:56:20 [NickH]
manu: even if we hate micro syntaxes, there are ways that we can hide them from the mainstream approach
14:56:34 [NickH]
webr3: +1
14:56:55 [Zakim]
14:56:58 [gavinc]
Do ANY of the Web Developer JSON syntax methods provide the BENEFITS of RDF to the JSON developer?
14:56:58 [NickH]
ivan: we should be optimising for web developers
14:57:13 [manu]
14:57:42 [NickH]
ivan: joking aside, being readable isn't as important as making it easy to use for web developers
14:57:44 [PatH]
+1 to speaker. Readers count less that developer code.
14:57:49 [PatH]
14:57:54 [NickH]
danbri: it is about code readability not data readability
14:57:55 [ivan]
ack manu
14:57:59 [gavinc]
14:58:00 [mischat]
14:58:09 [davidwood]
ack manu
14:58:11 [NickH]
manu: the only microformat we need is for languages, not for datatypes
14:58:12 [danbri]
s/not/as well as/ (well I can't remember what I said, but that's what I meant :)
14:58:16 [ivan]
s/to speaker/to Danbri/
14:58:26 [Zakim]
14:58:37 [webr3]
zakim, i am ??P7
14:58:37 [Zakim]
+webr3; got it
14:59:19 [pgroth]
manu, aren't you just saying, we should make this context with a bunch of default interpretations
15:00:00 [NickH]
manu: you can use the context to work out the type or datatype instead of microsyntaxes
15:00:18 [manu]
{"value": "5", "type": "xsd:integer"}
15:00:24 [NickH]
manu: for example when using foaf:homepage, you can guess that the value is an IRI
15:00:29 [manu]
{"value": "foo", "lang": "fr"}
15:00:40 [webr3]
or arcs..
15:00:49 [webr3]
I don't have much to say bar -> we can have two simple syntaxes, one v simple w/ no datatypes or langs, and another one just like jtriples w/ full rdf support, I see no need to complicate this - although we can and have a hybrid that covers every use cases, just worries me having one complex spec that can cover everything, vs two simple specs that people can grok in a few minutes / w/ one example
15:01:10 [ivan]
15:01:28 [ivan]
ack gavinc
15:01:32 [SteveH]
15:01:50 [sandro]
sandro: you could have foaf_name_en and foaf_name_gr etc....
15:02:10 [pgroth]
15:02:13 [pfps]
+1 to gavinc
15:02:25 [PatH]
Because they might need to interact with RDF data whether htey like it or not.
15:03:14 [NickH]
manu: it is about convicing the publishers that they don't need to change their JSON too much to make it RDF compatible
15:03:26 [webr3]
and I want to cover (in one case):
15:03:29 [manu]
q+ to discuss PaySwarm use case.
15:03:30 [cygri]
gavinc: possible answer here:
15:03:48 [manu]
15:03:57 [manu]
q+ to discuss Twitter/Facebook/etc.
15:04:07 [cygri]
15:04:12 [ivan]
15:04:21 [NickH]
manu: the target for the specification is not for developers, but for publishers
15:04:39 [sandro]
15:04:42 [webr3]
manu, example re facebook: <> If you somehow said "stick before each property, and use the URL as the subject, you've got RDF", then I believe that would cover a large portion of the "JSON as RDF" A/D user segments.
15:05:35 [sandro]
ack SteveH
15:05:39 [ivan]
ack SteveH
15:05:51 [NickH]
15:05:58 [NickH]
15:06:25 [manu]
What six syntaxes do they publish their API in?
15:06:26 [NickH]
SteveH: we don't want the JSON from the Twitter API to return RDF triples
15:06:32 [manu]
JSON, XML - what else?
15:06:33 [mischat]
15:06:33 [sandro]
SteveH: My concern is similar gavin's. I think there is very small audience for this. This doesnt provide anything anyone needs. Who wants the twitter API as JSON...? And Twitter is willing to produce other formats as well!
15:06:36 [pgroth]
15:06:43 [webr3]
manu, formerly atom and rss too iirc
15:06:51 [NickH]
SteveH: there just aren't enough consumers to justify the work
15:06:55 [sandro]
15:07:05 [manu]
webr3: but their API in Atom/RSS?
15:07:15 [Guus]
15:07:17 [webr3]
manu: previously ya, unsure now
15:07:20 [manu]
webr3: They publish /some/ data in Atom/RSS - but not all of it.
15:07:25 [yvesr]
15:07:35 [sandro]
SteveH: Also, on microsyntaxes -- if the data is not in RDF, how would get language tags, etc?
15:07:41 [NickH]
SteveH: if the data is not being held in RDF, I don't see why you need to encode the RDF datatypes into JSON, just use JSON types
15:07:42 [ivan]
ack manu
15:07:42 [Zakim]
manu, you wanted to discuss Twitter/Facebook/etc.
15:07:47 [sandro]
sandro: Manu is trying to address groups A, and B (and C) at the same time,
15:07:52 [davidwood]
ack manu
15:07:52 [danbri]
example from twitter: 'Supported formats XML, JSON'
15:08:01 [mischat]
15:08:38 [NickH]
manu: just to be clear, I don't suggest that we suggest to solve Groups A,B,C,D with a single solution, just that there are solutions on the table that are possible
15:08:56 [Guus]
I wonder wehther we should simply start from the proposals: only consider those proposals that already have a substantial user base
15:09:14 [mischat]
i wonder if developers use JSON because of good toolings, and simple code, as mentioned earlier, not because of the fact that the JSON syntax looks a certain way
15:09:21 [sandro]
(I don't think A and B can be satisfied together. There is no format which can be trivially parsed to triples and also is friendly to js app developers, for *all* RDF.)
15:09:25 [yvesr]
mischat: speed is a massive factor
15:09:37 [yvesr]
json parsers are often an order of magnitude faster than xml
15:09:57 [mischat]
indeed, and if the json parser had to understand rdf, i bet it would be less efficient
15:10:05 [sandro]
15:10:11 [Guus]
15:10:13 [danbri]
(with Facebook, the single biggest problem w/ RDFa is the disconnected namespace declaration --- publishers constantly screw that up)
15:10:15 [LeeF]
I don't think it's the WG's job to "place a bet" on what will get the rest of the world to adopt Semantic Web technologies.
15:10:23 [yvesr]
mischat: yes, you'd have to add a layer on the top casting that to a graph
15:10:32 [mischat]
15:11:32 [gavinc]
Yeah, I write 30 lines of Python ... and produce triples at the other end
15:11:40 [sandro]
steve: who would want to consume this stuff in preference to consumer normal json? We're an RDF shop, and we'd rather just consume JSON and turn in to RDF itself.
15:11:49 [MacTed]
rather frustrating to have the wiki say the next meeting is not until next week....
15:11:54 [sandro]
15:11:55 [davidwood]
ack cygri
15:12:17 [manu]
q+ to talk about decentralized systems
15:12:29 [pfps]
steve - is there a document that says how you process vanilla JSON and turn it into RDF
15:12:35 [sandro]
cygri: For sindice it would be great if it could just be indexing all these json data feeds without per-source coding.
15:12:35 [davidwood]
MacTed: When do you want the next meeting to be?
15:13:25 [mischat]
cygri: speak up
15:13:35 [mischat]
15:14:19 [NickH]
cygri: do we have anything to offer in the Group A just care about JSON and don't actually care about RDF? Yes, we do. RDF has a number of advtanages as a data model.
15:14:25 [yvesr]
15:14:26 [SteveH]
cygri's point is reasonably persuasive, but were a few hops from that in the tech world curently
15:14:30 [NickH]
cygri: in RDF, URIs are explicitly marked up
15:14:33 [webr3]
question: if you took jtriples or talis-rdf as one serialization (to cover rdf), and created a way for the opengraph data to say "append the property names to and use the GET uri as a subject" (to bring linked data/rdf basic benefits to the wild web), then what segment would not be covered?
15:14:35 [webr3]
.. also, if JSON-schema was merged w/ owl in some way, surely that'd cover everything possible?
15:14:44 [mischat]
15:14:53 [NickH]
cygri: terms in the data dictionary are unique and resolvable
15:15:03 [NickH]
cygri: in RDF it is easy to mix data dictionaries
15:15:27 [NickH]
cygri: there are number of properties that RDF has that JSON doesn't have, that can benifits the JSON community
15:15:56 [danbri]
(seems to be some discussion of RDF on the JSON schemas list - )
15:15:59 [NickH]
cygri: it could make JSON a tiny bit better without implementing full RDF
15:16:40 [cygri]
the mail i mentioned is here:
15:16:42 [davidwood]
ack ivan
15:16:43 [sandro]
cygri: These has to be a benefit to JSON producers and JSON consumers to this work, or it's not worth doing.
15:16:50 [webr3]
danbri, I've been talking to kris zyp too, and he's def got interest there, I've offered to help mix in benefits of rdf/owl/linked-data to json-schema
15:17:10 [NickH]
ivan: the semantic web community has produced huge amount of data on the web
15:17:38 [mischat]
+1 to ivan's point. I went to the UK government hack day, and NOT ONE of the web developers uses any of the RDF on
15:17:42 [manu]
+1 to Ivan!!
15:17:46 [pgroth]
15:17:48 [mischat]
they ONLY used the JSON data
15:17:48 [sandro]
ivan: The problem is that the web developers ignore what the LOD community has produced.
15:17:50 [NickH]
ivan: there is a large community of web mashup developers out there that today ignores the data that the linked open data community produces
15:17:54 [danbri]
q+ to suggest that consuming apps and useful tooling are more important than syntax tweaks
15:18:01 [yvesr]
althouth the web dev community likes linked data when it's *also* available as JSON
15:18:08 [yvesr]
we have quite a lot of people using our BBC JSON feeds
15:18:09 [cygri]
15:18:10 [NickH]
ivan: today they are unwilling to go outside of their world and use Turtle
15:18:25 [sandro]
sandro: Why doesn't the SPARQL JSON Result format work for them, Ivan?
15:18:28 [sandro]
ivan: It might.
15:18:29 [pgroth]
what we need, is if I expose my data as linked data, I get for free a json api
15:18:32 [NickH]
ivan: that they need, is very clearly, a simple JSON view on the data
15:18:41 [PatH]
+1 to pgroth
15:18:47 [pchampin]
what about a JSON-CONSTRUCT keyword in SPARQL, then?
15:18:59 [yvesr]
pgroth, yes, exactly
15:19:03 [davidwood]
ack guus
15:19:14 [NickH]
davidwood: I agree that they are two completely different communities
15:19:23 [yvesr]
pgroth, the inverse transformation is still interesting though
15:19:31 [PatH]
SOunds like the most urgent JSON need is for a simple API for JSON developers, to lure them into using RDF. Focus on this?
15:19:49 [pgroth]
PatH, agree
15:20:08 [pgroth]
what about the RDFapi group for this?
15:20:31 [zwu2]
If people need to use rdf, they use it. If there is no real business need, they won't.
15:20:32 [sandro]
Yes, Pat -- that would address Ivan's needs, probably. A very nice rdf.js.
15:20:39 [yvesr]
PatH, agree
15:20:52 [NickH]
Guus: I don't see making simple JSON based API for this group. Still after two months we have not made any progress.
15:20:54 [sandro]
15:20:58 [davidwood]
ack manu
15:20:58 [Zakim]
manu, you wanted to talk about decentralized systems
15:21:02 [LeeF]
zwu2++ -- if people don't need to use RDF and are happy with JSON, then there's no problem to be solved there
15:21:09 [SteveH]
15:21:44 [danbri]
15:21:51 [NickH]
manu: at one point it was asked "What can we offer Group A?". The JSON community doesn't current have a way to build de-centralised systems.
15:22:11 [MacTed]
davidwood - the "next meeting" is apparently today.
15:22:12 [NickH]
manu: at the moment you have to re-invent the wheel every time you want to talk to a different service
15:22:40 [NickH]
manu: there is a need to have a de-centralised, simple, communication protocol
15:22:46 [davidwood]
MacTed: We are at F2F1 now, but if you want to change the wiki you certainly may.
15:22:58 [Zakim]
+ +1.781.273.aaii
15:22:59 [ivan]
15:23:00 [davidwood]
15:23:08 [NickH]
Guus: is this better suited to the RDF API group?
15:23:08 [davidwood]
ack mischat
15:23:15 [ivan]
15:23:20 [danbri]
15:23:51 [NickH]
mischat: The UK Government has lots of data. Recently went to a Hack the Government day. Lots of good developers but none of them know how to use RDF.
15:23:57 [danbri]
15:23:58 [NickH]
mischat: JSON offers simplicity
15:24:00 [LeeF]
That sounds like success to me. RDF was valuable to bring the government data together (right?), and yet didn't have to cause anyone to change their toolchains to consume the data. That's a good thing, right?
15:24:06 [pchampin]
q+ to ask if we can bring the benefits of RDF to people that will stick to JSON objects
15:24:11 [davidwood]
ack cygri
15:25:00 [NickH]
cygri: we have 3 distinct problems that we are trying to solve. And we need to handle them seperately.
15:25:13 [NickH]
cygri: 1) we need to make simple JSON more RDFy
15:25:24 [pchampin]
s/simple/existing simple/
15:25:54 [NickH]
cygri: 2) If we already have RDF on the publisher side, we should make it available as JSON for JavaScript consumers
15:25:55 [mischat]
15:25:58 [webr3]
there is a huge quantity of non rdf data in the world, and a lot of that in json, a way to see that data as simple rdf and mash it up with data from other non rdf sources would be, well huge imho - easiest way is shared property names and uris as ids
15:26:07 [sandro]
cygri: three distinct problems. we have to separate them. (1) trying to get data out of current JSON APIs, making them more RDFy. (2) if we have data in RDF at the publisher and the consumer wants RDF. Eg for SPARQL Construct result in JS. rdf-2-rdf. (3) You have data in RDF and you want web developers to use this..
15:26:14 [sandro]
+1 cygri
15:26:19 [davidwood]
15:26:20 [SteveH]
15:26:26 [NickH]
cygri: 3) Developers don't need to know anything about RDF, just make it easy for them to consume data that is already in RDF
15:26:37 [webr3]
1 = 3 ?
15:26:38 [davidwood]
ack ivan
15:26:40 [SteveH]
q+ to talk about JSON APIs to RDF (3)
15:26:43 [ivan]
ack ivan
15:27:04 [danbri]
some numbers:
15:27:13 [cygri]
webr3: no, not at all. in 1, the publisher has json and the consumer wants rdf. in 3, the publisher has rdf and the consumer wants json
15:27:25 [NickH]
ivan: comment on the API stuff. What the other working group is working on is a JavaScript API for RDF people.
15:27:37 [webr3]
cygri, yes, but the solution for one, serializationwise, would suit both 1 and 3
15:28:05 [NickH]
ivan: the type of API that you see in Jena or rdflib in Python is being implemented in JavaScript for RDF savy people
15:28:10 [Zakim]
15:28:34 [Zakim]
15:28:40 [webr3]
zakim, i am IPcaller
15:28:40 [Zakim]
ok, webr3, I now associate you with [IPcaller]
15:28:57 [NickH]
danbri: :(
15:29:06 [davidwood]
ack pchampin
15:29:06 [Zakim]
pchampin, you wanted to ask if we can bring the benefits of RDF to people that will stick to JSON objects
15:29:08 [cygri]
webr3, you keep saying that but i don't believe it
15:29:26 [sandro]
-1 I think a *good* RDF JS API would allow web developers to get at RDF data.
15:29:42 [davidwood]
I think Web developers want useful data, not to learn RDF.
15:29:44 [sandro]
That is: -1 Ivan, because I think a *good* RDF JS API would allow web developers to get at RDF data.
15:29:59 [webr3]
cygri, can you think of one feature that oen would have that the other would not?
15:30:03 [manu]
Sandro - I don't think that's what Ivan was saying? Or I don't understand your response...
15:30:08 [davidwood]
ack SteveH
15:30:08 [Zakim]
SteveH, you wanted to talk about JSON APIs to RDF (3)
15:30:10 [NickH]
pchampin: seperate out the problems
15:30:52 [cygri]
webr3, the 1 solution needs uris, the 3 solution doesnt
15:31:13 [webr3]
cygri, uri's for properties?
15:31:22 [sandro]
steve: To expose RDF data to web developer, pre-write some canned SPARQL queries and serialize the output as native engineered JSON.
15:31:27 [NickH]
SteveH: provide pre-canned SPARQL queries to expose JSON to web developers
15:31:29 [cygri]
webr3 for properties and instances
15:31:42 [sandro]
steve: No need for stds here.
15:31:54 [NickH]
SteveH: give them a data format that web developers want to consume - give them pure JSON, not RDF disguised as JSON
15:32:32 [NickH]
ivan: eveything that Linked Data community is doing is being ignored
15:32:34 [webr3]
cygri, so if there was a way to say "append to all properties and use the GET URI as a subject" in both solutions, what would the difference be?
15:32:48 [cygri]
webr3, that doesn't work for 3
15:33:07 [webr3]
cygri, why not, the uri isn't in the data in both cases..
15:33:18 [sandro]
15:33:20 [ivan]
15:33:47 [cygri]
webr3 maybe i don't understand what you mean by "there is a way"
15:33:53 [SteveH]
15:34:09 [NickH]
danbri: there is all this beautiful data being pubished by the linked data world. But it isn't being used. But I don't think a new standard is the way to solve this.
15:34:32 [LeeF]
SWEO was the right place for this sort of thing. Not the RDF WG.
15:34:35 [NickH]
ivan: nobody is doing anything significant to get out of the chicken and egg problem
15:34:51 [webr3]
cygri, it'd just be like perhaps with one addition { vocab: } in the data
15:35:07 [danbri]
15:35:12 [webr3]
cygri, although i can't see any reason for that not to be in a json-schema like doc..
15:35:30 [davidwood]
ack SteveH
15:35:32 [danbri]
15:35:36 [danbri]
15:35:55 [pgroth]
15:36:08 [NickH]
yvesr: the BBC offers both RDF and JSON but most people are consuming the JSON, including danbri
15:36:33 [NickH]
yvesr: the JSON is the same model as the RDF with the namespaces stripped out - very simple
15:36:52 [mischat]
ivan: this is a pretty printed version of the BBC json :
15:37:32 [danbri]
this google sgapi eats rdf/xml with a real (raptor) parser -
15:37:45 [NickH]
ivan: what are the real big success stories of linked data?
15:37:57 [danbri]
(ie. every FOAF file they found in the Web, via standard Google Web crawl)
15:38:06 [davidwood]
ack pgroth
15:38:11 [manu]
q+ on needing a publisher in the middle
15:38:23 [PatH]
Seems to me that I keep hearing that people want to use JSON because it is simple, and not use RDF because it is complicated. In which case, a complicated embedding of full RDF into JSON seems like a shot in the foot.
15:38:33 [NickH]
davidwood: how do we take the linked data cloud and make it available to web developers, without something re-publishing it in the middle
15:38:40 [webr3]
.. ivan,all publish my data from bill roberts does that..
15:38:42 [sandro]
(agreed, PatH.)
15:38:46 [manu]
15:39:00 [mischat]
+1 to PatH
15:39:04 [manu]
q+ to talk about democratization of data and not needing SPARQL publishers.
15:39:33 [pgroth]
15:39:37 [NickH]
pgroth: there just needs to be recipies for publishing RDF as simple JSON
15:39:39 [davidwood]
15:40:12 [PatH]
Being able to just see Paul's right hand is kind of amusing.
15:40:31 [PatH]
15:40:59 [gavinc]
So extend it! And if someone uses it!
15:41:15 [danbri]
so is a good example; they do publish RDFa in various vocabs. Invalid markup at various other levels, though.
15:41:24 [danbri]
q+ to work through slideshare situation
15:41:30 [NickH]
SteveH: what would help is if you could provide a JSON template for exporting RDF as JSON using SPARQL
15:41:43 [davidwood]
ack manu
15:41:43 [Zakim]
manu, you wanted to talk about democratization of data and not needing SPARQL publishers.
15:41:47 [webr3]
PatH +10 to that
15:42:09 [danbri]
('complicated' is a complicated notion)
15:42:24 [NickH]
manu: havn't a publisher in the middle is a fail stategy for Linked Data
15:42:29 [SteveH]
like CONSTRUCT JSON { { "name": ?name, "dob":?dob } } WHERE { ?x :name ?name ; :dob ?dob }
15:42:31 [SteveH]
or whatever
15:42:42 [davidwood]
+1 to manu
15:42:42 [sandro]
-1 manu. Actually, I think having services which gateway RDF to WebApps is okay.
15:42:44 [pfps]
+1 to manu
15:42:53 [SteveH]
-1 to manu
15:42:53 [pgroth]
popular web technologies provide easy reciepes
15:42:58 [SteveH]
15:42:59 [PatH]
danbri, it all depends on what 'is' is.
15:43:09 [webr3]
pgroth, +1 easy recipes
15:43:29 [cygri]
15:43:46 [webr3]
-1 we should be addressing both, seperately - and can, easily
15:43:54 [NickH]
manu: we do want people to be currating the data, but we should focus on the Web Developers not on the few people with a big triplestore
15:43:54 [sandro]
+1 to manu: we should be addressing (small) web developers
15:44:28 [sandro]
I'm not suggesting the gateways do any curating.
15:45:12 [pgroth]
manu, you want to just only use the json pipeline for rdf?
15:45:17 [PatH]
15:45:20 [davidwood]
ack danbri
15:45:20 [Zakim]
danbri, you wanted to work through slideshare situation
15:45:29 [SteveH]
15:45:48 [danbri]
15:46:01 [danbri]
15:46:41 [webr3]
following nose /SHOULD/ be huge, many people wants shared schema's between domains and the ability to look up id's easily, it just isn't hugely popular because RDF is way too complicated for most, and linked data currently focuses a lot of time on RDF, or SPARQL
15:46:45 [davidwood]
ack SteveH
15:47:01 [NickH]
danbri: SlideShare do publish RDF but they are using RDFa, not JSON
15:47:06 [sandro]
15:47:11 [mischat]
here is more triples when you tell rapper it that the file contains rdfa :
15:47:12 [davidwood]
ack cygri
15:47:27 [manu]
pgroth: Yes.
15:48:30 [danbri]
15:49:18 [sandro]
(I'm starting to reluctantly agree with Richard.)
15:49:20 [NickH]
cygri: solving the problem of getting lots of developers to consume Linked Data is not a job for a standardisation group
15:49:25 [cmatheus]
15:49:29 [SteveH]
15:49:35 [sandro]
15:49:38 [danbri]
well, except if millions of otherwise smart developers are ignoring our tech, it might be worth exploring why
15:49:46 [NickH]
Guus: we can start with a minimum solution
15:49:50 [LeeF]
cygri +1
15:49:57 [PatH]
+1. KISS.
15:50:05 [yvesr]
not sure i agree, standards should be made to be useful to a large extent...
15:50:06 [zwu2]
+1 to a minimum solution
15:50:10 [webr3]
+1, simple as possible
15:50:13 [PatH]
(another Clinton quote, FWIW)
15:50:20 [manu]
+1 to a minimal solution
15:50:21 [yvesr]
where useful means a large number of devs using them
15:51:02 [NickH]
cygri: I would peronally like to see these problems solved but I am not convinced that we have a good solution for peoplems number 1 and 2
15:51:05 [NickH]
cygri: I would peronally like to see these problems solved but I am not convinced that we have a good solution for peoplems number 1 and 3
15:51:12 [danbri]
semweb = standards + community + tools + data; if the outside world is ignoring, I'd suggest refocussing as semweb = data + tools + community + standards. The data's the prize, everything else is a means to an end.
15:51:34 [yvesr]
danbri, +1
15:51:39 [danbri]
jsonld =
15:51:42 [webr3]
danbri, +1
15:51:44 [danbri]
'JSON-LD - Expressing Linked Data in JSON '
15:51:46 [PatH]
danbri, +10
15:51:46 [NickH]
1: JSON-LD object style.
15:51:50 [sandro]
15:51:58 [NickH]
2: Talis JSON/RDF
15:52:16 [NickH]
3: Linked Data API - we have triples but want to expose them as JSON
15:52:53 [danbri]
I don't think most JSON enthusiasts care about NOTE vs REC
15:53:04 [mischat]
+1 to danbri
15:53:12 [NickH]
cygri: problem 2 is fairly clear how to solve
15:54:08 [PatH]
'Note' is a black hole to drop overenthusiastic WG ideas into.
15:54:31 [PatH]
15:54:32 [cygri]
PathH lol
15:54:41 [NickH]
15:54:44 [danbri]
zakim, who is ringing?
15:54:44 [Zakim]
I don't understand your question, danbri.
15:54:49 [danbri]
zakim, who is speaking?
15:54:49 [manu]
zakim, who is making noise?
15:54:59 [Zakim]
danbri, listening for 10 seconds I heard sound from the following: Meeting_Room (45%), zwu2 (10%), OlivierCorby.aa (54%), +1.781.273.aaii (54%)
15:55:09 [Zakim]
manu, listening for 10 seconds I heard sound from the following: gavinc (9%), Meeting_Room (50%), OlivierCorby.aa (40%), +1.781.273.aaii (15%)
15:55:43 [MacTed]
Zakim, aaii is OpenLink_Software
15:55:43 [Zakim]
+OpenLink_Software; got it
15:55:43 [MacTed]
zakim, OpenLink_Software is temporarily me
15:55:43 [MacTed]
zakim, mute me
15:55:44 [Zakim]
+MacTed; got it
15:55:44 [Zakim]
MacTed should now be muted
15:55:55 [zwu2]
zakim, mute me
15:55:55 [Zakim]
zwu2 should now be muted
15:57:07 [manu]
I think guidance isn't going to do anything significant to change Linked Data's position in the world.
15:57:09 [danbri]
q+ to ask about json schemas
15:57:26 [NickH]
davidwood: if we are going to give up on developers in terms of devloping a standard for consuming RDF as JSON
15:57:46 [NickH]
SteveH: I don't like using the phrase 'giving up on developers'
15:57:58 [manu]
I think that's exactly what we're doing - "Giving up on developers"
15:57:58 [danbri]
15:58:01 [webr3]
danbri, thanks.. json-schema + owl2 == success imo
15:58:21 [danbri]
webr3, i didn't follow all the list - did it get any discussion in this wg yet?
15:58:25 [SteveH]
manu, where are these developers that want to consume JSON as RDF? I don't know any
15:58:37 [NickH]
davidwood: we can still help developers by offering advice, rather than creating a standard
15:58:45 [SteveH]
manu, ah, except cygri's Sindice usecase
15:58:46 [pchampin]
@web3: just a pronostic, or have you any experience on that?
15:58:48 [webr3]
danbri, not really, I've mentioned it a few times, pushback or nothing was the response
15:58:58 [SteveH]
but they're RDF developers, not "web" developers
15:59:08 [manu]
SteveH: You know one company - that's us. PaySwarm is has these requirements coming down the pike very soon, which will also need this.
15:59:30 [SteveH]
manu, ok, you'll have to explain why the triples help, offline
15:59:37 [PatH]
+1 to speaker
15:59:50 [manu]
SteveH: Yes, I think we'll have to have a phone chat at some point because I think we're speaking past each other.
15:59:55 [sandro]
+1 DavidWood: Do (2), Do (3) if someone wants to, and (1) if a good proposal emerges.
15:59:58 [webr3]
pchampin, owl community are sayign it's a good thing, json schema community is saying it'd be good, just needs to happen (and it seems pretty obvious)
16:00:10 [SteveH]
manu, yeah, I think so too
16:00:26 [manu]
Do you want to setup a time now? I have time on Friday - or early next week.
16:01:21 [SteveH]
manu, I can do monday afternoon, uk time
16:01:28 [webr3]
SteveH, I'm also one, and so are most of the dev's i work with on 3 different non-rdf projects, they all want the simple benefits
16:01:46 [danbri]
so re schema-annotation/grddl-ish approach, ... i hear luke-warm mild curiousity, but no huge enthuasism here yet
16:01:56 [NickH]
Guus: Proposal to do work on Case Type 2, starting with the Talis JSON/RDF
16:02:13 [SteveH]
webr3, but, like danbri said, consuming RDF is a pain, you really need a triplestore (even if in memory)
16:02:25 [PatH]
In favor.
16:02:50 [PatH]
Woot, a proposal to do something.
16:02:56 [NickH]
Guus: Proposal to do work on Case Type 2, starting with the Talis JSON/RDF. Revisit 1 in the future
16:03:30 [webr3]
danbri, see (half way down)
16:04:22 [webr3]
deiter(?) it doesn't - it's just for a fast light over the wire RDF serialization
16:04:29 [sandro]
PROPOSED: (1) Incubate on something like JSON-LD, (2) make a REC on something like Talis RDF/JSON, and (3) make a Note on current practice stuff like Linked Data API.
16:04:45 [sandro]
(got 12 +1's in the room)
16:04:50 [webr3]
16:05:01 [manu]
wait what? Did we straw-poll already?
16:05:14 [PatH]
16:05:15 [manu]
16:05:21 [davidwood]
Manu: Please vote note
16:05:24 [zwu2]
16:05:25 [davidwood]
16:05:33 [LeeF]
16:05:37 [gavinc]
16:05:46 [PatH]
16:05:53 [davidwood]
Manu: Is that a formal objection?
16:05:56 [gavinc]
(It's what we can do, not what we should do)
16:05:57 [danbri]
can we try an exercise: compare sample client code for with a triples JSON syntax
16:05:57 [LeeF]
Can we get the straw poll feeling of the people in the room recorded in the minutes, please?
16:06:16 [manu]
No, not a formal objection - there's much teeth grinding, but no formal objection :)
16:06:49 [webr3]
manu, tis better than a year of fighting because there are different use cases, background, needs etc - at least there's a chance of two decent specs at the end this way
16:06:56 [manu]
but I am a very strong -1 - I think this group is focusing on the wrong thing by focusing on the group that's already sold on using RDF/TURTLE/SPARQL/etc.
16:07:08 [sandro]
LeeF, as I said, "(got 12 +1's in the room) " trying to characterize the room.
16:07:19 [LeeF]
sandro, I don't know who that is though
16:07:44 [LeeF]
thank you!
16:07:56 [sandro]
PROPOSED: (1) Incubate on something like JSON-LD, (2) make a REC on something like Talis RDF/JSON, and (3) make a Note on current practice stuff like Linked Data API.
16:07:59 [webr3]
1st: can you define "incubate on"
16:08:00 [sandro]
16:08:02 [SteveH]
16:08:03 [davidwood]
16:08:04 [pchampin]
16:08:06 [LeeF]
16:08:07 [mbrunati]
16:08:07 [zwu2]
16:08:08 [NickH]
16:08:10 [PatH]
16:08:11 [cmatheus]
16:08:12 [ivan]
16:08:12 [gavinc]
16:08:14 [FabGandon]
16:08:18 [JFB]
16:08:25 [cygri]
16:08:26 [manu]
-1 I think this group is focusing on the wrong thing by focusing on the group that's already sold on using RDF/TURTLE/SPARQL/etc.
16:08:28 [danbri]
manu, do you have any interest to work on json-schema-based approach?
16:08:31 [mischat]
16:08:35 [danbri]
16:08:37 [webr3]
16:08:37 [NickH]
sandro: it means if a good proposal comes back in a few months time, then we can do something about it
16:08:45 [manu]
danbri: Perhaps - it's something we talked about internally
16:08:57 [webr3]
danbri, manu, i really do, so does kris zip who does json schema
16:09:01 [sandro]
sandro: "incubate on" means not spend serious WG time, but we're free to revisit it later in the life of this WG and maybe adopt it.
16:09:03 [cmatheus]
this sounds like it may be the only pragmatic way to proceed but to me it doesn't seem to be going far enough unless we eventully get back to something along the lines of JSON-LD
16:09:04 [NickH]
sandro: but not spend any more working group time on it for the time being
16:09:07 [danbri]
action: danbri follow up with manu regarding schema-based mapping of json into rdf
16:09:08 [trackbot]
Created ACTION-31 - Follow up with manu regarding schema-based mapping of json into rdf [on Dan Brickley - due 2011-04-20].
16:09:31 [NickH]
davidwood: I don't think there is something useful we can do for the comunity for the time being
16:09:46 [pfps]
16:09:48 [PatH]
what is the difference between +0 and -0 ?
16:10:15 [manu]
PatH: +0 - you are smiling while you do it, -0 you're frowning :)
16:10:21 [PatH]
16:10:22 [gavinc]
-0: 'I won't get in the way, but I'd rather we didn't do this.'
16:10:27 [gavinc]
+0: 'I don't feel strongly about it, but I'm okay with this.'
16:10:28 [manu]
+1 to pfps
16:10:43 [webr3]
+1 to pfps as well
16:10:49 [JFB]
Then I change my vote from +0 to -0
16:11:36 [cygri]
16:11:49 [davidwood]
webr3: That's what "incubate" means :)
16:12:06 [webr3]
davidwood, has been incubating with some for 1 year+ already
16:12:07 [PatH]
Peter's concern is worth noting. If the WG can un-incubate this later, that would be a Good Thing. We should not just let it die.
16:12:46 [cygri]
manu +1
16:12:47 [davidwood]
I agree that the WG should revisit
16:13:41 [pchampin]
16:13:50 [Zakim]
16:14:09 [davidwood]
ack cygri
16:14:16 [FabGandon]
FabGandon has left #rdf-wg
16:14:50 [FabGandon]
FabGandon has joined #rdf-wg
16:15:30 [NickH]
cygri: a lot of the JSON discussion as been about the differences between approach 1 and approach 2. Aprooach 3 has hardly been looked at so far. Can the working group look at approach 3?
16:16:31 [danbri]
three versions of the BBC program description from yvesr & co:
16:16:34 [AZ]
AZ has joined #rdf-wg
16:16:58 [PatH]
Sandro, relax. You are making me dizzy.
16:17:06 [sandro]
16:17:09 [PatH]
BUt thanks.
16:17:24 [PatH]
16:17:58 [PatH]
is of is not?
16:18:05 [PatH]
16:18:18 [webr3]
yvesr, it's not it's "RDF Web Applications WG"
16:19:04 [Zakim]
16:19:13 [AZ]
zakim, mute me
16:19:13 [Zakim]
AZ should now be muted
16:20:18 [sandro]
RESOLVED: (1) Incubate on something like JSON-LD, (2) make a REC on something like Talis RDF/JSON, and (3) make a Note on current practice stuff like Linked Data API.
16:20:40 [NickH]
manu: I can't accept the decision and plan to go and talk to JSON providers and see what will work with them
16:21:07 [sandro]
16:21:08 [danbri]
fyi json-ld list is at
16:21:24 [NickH]
mischat: oops
16:21:38 [NickH]
manu: I can accept the decision and plan to go and talk to JSON providers and see what will work with them
16:21:54 [sandro]
discussing possible public-rdf-json mailing list
16:22:22 [webr3]
which group would it be associated with?
16:22:32 [webr3]
(or none)
16:22:36 [sandro]
ACTION: sandro to make
16:22:36 [trackbot]
Created ACTION-32 - Make [on Sandro Hawke - due 2011-04-20].
16:22:54 [Zakim]
16:22:58 [zwu2]
bye, time to sleep now.
16:23:02 [Zakim]
16:23:05 [NickH]
davidwood: we should create a W3C mailing list for investigation into RDF JSON
16:23:29 [sandro]
manu, or maybe the list should be called something else, to separate it as this kind of JSON....?
16:23:36 [NickH]
guus: brainstorm - what breakout topics should we go for
16:23:43 [NickH]
16:23:48 [sandro]
public-rdf-view-of-json :-)
16:23:58 [webr3]
@sandro, could the list also be used for the owl+json-schema possible work/discussions that are going on too? (re the name)
16:24:19 [sandro]
I think so.
16:24:23 [manu]
I'm concerned about having 'rdf' in the name of the mailing list :)
16:24:30 [webr3]
16:24:33 [manu]
16:24:34 [PatH]
name, rjdsfon
16:24:44 [manu]
16:24:57 [sandro]
Yes..... I like that.
16:25:48 [webr3]
16:26:08 [sandro]
question whether it's a task force of this WG and still covered by the W3C patent policy or not.
16:26:56 [sandro]
in-wg gets patent policy; out-of-wg gets everyone to join.
16:27:07 [sandro]
No, no one is scribing. Trying to figuoure out breakouts.
16:27:12 [webr3]
@sandro, community group for it?
16:27:42 [gavinc]
Which break outs will have a phone?
16:28:04 [sandro]
zakim, who is talking?
16:28:04 [mischat]
zakim, who is making noise ?
16:28:13 [PatH]
Sounds like wqe are now a fax machine
16:28:15 [Zakim]
sandro, listening for 10 seconds I heard sound from the following: gavinc (9%), Meeting_Room (54%)
16:28:26 [Zakim]
mischat, listening for 10 seconds I heard sound from the following: gavinc (35%), Meeting_Room (15%)
16:28:44 [manu]
I have to run - thanks for the meeting all - sorry I couldn't be there F2F :)
16:29:21 [mischat]
bye manu
16:29:23 [sandro]
guus: Start tomorrow with 2 hours on 4 critical issues on GRAPHS.
16:29:40 [PatH]
What time is that 'start' tomorrow?
16:30:30 [webr3]
will skolem breakout be on zakim?
16:30:31 [pfps]
9:30 Amstertam
16:30:45 [sandro]
no clear we'll do the Skolem breakout.
16:30:57 [FabGandon]
FabGandon has left #rdf-wg
16:31:07 [PatH]
So just to be clear, 9.30 Amsterdam for the 2-houir on graph ccritical issues?
16:31:26 [AZ]
enjoy your dinner, bye
16:31:34 [Zakim]
16:31:39 [PatH]
Enjoy your dinner and walks.
16:31:43 [Zakim]
16:32:31 [gavinc]
Okay, see everyone in the morning.
16:32:39 [Zakim]
16:32:52 [Zakim]
16:32:53 [Zakim]
16:33:35 [Zakim]
16:34:57 [Zakim]
16:39:10 [Zakim]
- +1.603.897.aahh
18:56:43 [tomayac]
tomayac has joined #rdf-wg
20:32:54 [Scott]
Scott has joined #rdf-wg
20:45:32 [Scott]
Scott has joined #rdf-wg
20:49:20 [tomlurge]
tomlurge has left #rdf-wg
20:54:28 [tomlurge]
tomlurge has joined #rdf-wg
21:00:28 [cmatheus]
cmatheus has joined #rdf-wg
21:24:00 [JFB]
JFB has joined #rdf-wg
21:27:36 [JFB]
JFB has joined #rdf-wg
21:56:38 [pgroth]
pgroth has joined #rdf-wg
21:57:43 [SteveH]
SteveH has joined #rdf-wg
22:09:04 [mischat]
mischat has joined #rdf-wg
22:38:41 [danbri]
danbri has joined #rdf-wg
23:47:37 [LeeF]
LeeF has joined #rdf-wg