IRC log of sparql on 2010-07-20

Timestamps are in UTC.

13:54:57 [RRSAgent]
RRSAgent has joined #sparql
13:54:57 [RRSAgent]
logging to http://www.w3.org/2010/07/20-sparql-irc
13:54:59 [Zakim]
Zakim has joined #sparql
13:55:03 [sandro]
zakim, this will be sparql
13:55:05 [Zakim]
ok, sandro; I see SW_(SPARQL)10:00AM scheduled to start in 5 minutes
13:57:29 [kasei]
sandro, did you have thoughts on my sd:name followup email?
13:58:35 [AndyS]
Problems with zakim? I get junk sounds back when I dial in.
13:58:38 [sandro]
yes, kasei but I didn't finish sending them. :-(
13:58:42 [MattPerry]
MattPerry has joined #sparql
13:58:45 [kasei]
ok
13:59:07 [AxelPolleres]
AxelPolleres has joined #sparql
13:59:17 [sandro]
short version: those aren't problems -- I stand by my proposal.
13:59:27 [LeeF]
RRSAgent, make logs world
13:59:28 [Zakim]
SW_(SPARQL)10:00AM has now started
13:59:35 [Zakim]
+ +1.603.897.aaaa
13:59:40 [Zakim]
+Sandro
13:59:42 [Zakim]
- +1.603.897.aaaa
13:59:44 [Zakim]
+[IPcaller]
14:00:00 [kasei]
Zakim, [IPcaller] is me
14:00:00 [Zakim]
+kasei; got it
14:00:04 [bglimm]
I also get all circuitsare busy now when calling
14:00:13 [Zakim]
+pgearon
14:00:23 [Zakim]
+ +33.4.92.38.aabb
14:00:23 [NickH]
NickH has joined #sparql
14:00:24 [Zakim]
+ +1.603.897.aacc
14:00:25 [sandro]
AndyS, bglimm -- are you using the UK number or the US number?
14:00:26 [Zakim]
+Lee_Feigenbaum
14:00:30 [sandro]
zakim, what is the code?
14:00:30 [Zakim]
the conference code is 77277 (tel:+1.617.761.6200 tel:+33.4.26.46.79.03 tel:+44.203.318.0479), sandro
14:00:31 [MattPerry]
zakim, aacc is me
14:00:32 [Zakim]
+MattPerry; got it
14:00:34 [Zakim]
+AxelPolleres
14:00:35 [kasei]
ack [IPcaller]
14:00:41 [bglimm]
I am trying the new UK number
14:00:47 [OlivierCorby]
zakim, aabb is me
14:00:47 [Zakim]
+OlivierCorby; got it
14:00:55 [AxelPolleres]
Zakim, who is on the phone?
14:00:55 [Zakim]
On the phone I see Sandro, kasei, pgearon, OlivierCorby, MattPerry, Lee_Feigenbaum, AxelPolleres
14:00:55 [sandro]
I know the france number was not working yesterday. :-(
14:01:02 [AndyS]
I get invalid UK number
14:01:22 [bglimm]
I am trying France now
14:01:35 [ivan]
zakim, dial ivan-voip
14:01:35 [Zakim]
ok, ivan; the call is being made
14:01:37 [Zakim]
+Ivan
14:01:50 [bglimm]
Zakim, passcode
14:01:50 [Zakim]
I don't understand 'passcode', bglimm
14:02:04 [sandro]
Zakim, what is the passcode?
14:02:04 [Zakim]
the conference code is 77277 (tel:+1.617.761.6200 tel:+33.4.26.46.79.03 tel:+44.203.318.0479), sandro
14:02:13 [AxelPolleres]
andy, did you use the new number?
14:02:13 [Zakim]
+ +44.186.528.aadd
14:02:15 [Zakim]
+[IPcaller]
14:02:22 [bglimm]
Zakim, +44.186.528.aadd is me
14:02:22 [Zakim]
+bglimm; got it
14:02:26 [AndyS]
Yes - used the new number.
14:02:28 [bglimm]
France worked now
14:02:34 [bglimm]
UK doesn't
14:02:49 [AndyS]
zakim, [IPCaller] is me
14:02:49 [Zakim]
+AndyS; got it
14:02:54 [kasei]
Zakim, who is talking?
14:02:57 [Zakim]
+ +1.603.897.aaee
14:02:57 [bglimm]
Zakim, mute me
14:02:58 [Zakim]
bglimm should now be muted
14:03:05 [Zakim]
kasei, listening for 10 seconds I heard sound from the following: AxelPolleres (38%), MattPerry (10%), AndyS (42%), Ivan (15%)
14:03:08 [sandro]
I hear: <amy> i believe that the new zakim # is limited to a smaller number it can handle at a time. people should keep trying slowly
14:03:12 [bglimm]
Zakim, unmute me
14:03:12 [Zakim]
bglimm should no longer be muted
14:03:15 [chimezie]
chimezie has joined #sparql
14:03:19 [Souri]
Souri has joined #sparql
14:03:25 [chimezie]
Zakim, what is the passcode?
14:03:25 [Zakim]
the conference code is 77277 (tel:+1.617.761.6200 tel:+33.4.26.46.79.03 tel:+44.203.318.0479), chimezie
14:03:34 [AxelPolleres]
Zakim, who is on the phone?
14:03:34 [Zakim]
On the phone I see Sandro, kasei, pgearon, OlivierCorby, MattPerry, Lee_Feigenbaum, AxelPolleres, Ivan, bglimm, AndyS, +1.603.897.aaee
14:03:58 [Zakim]
+Chimezie_Ogbuji
14:04:10 [NickH]
NickH has joined #sparql
14:04:40 [AxelPolleres]
nick, do you plan to dial in, can you scribe?
14:04:56 [NickH]
yeah, tryiing to dial in, but the UK number isn't working for me
14:05:06 [AxelPolleres]
scribe: ChimezieOgbuji
14:05:13 [AxelPolleres]
chair: AxelPolleres
14:05:17 [NickH]
and had to kick someone out of the room I booked :(
14:05:21 [AxelPolleres]
regrets: AlexandrePassant
14:05:35 [AxelPolleres]
agenda: http://www.w3.org/2009/sparql/wiki/Agenda-2010-07-20
14:05:56 [chimezie]
Axel: back from vacation, still catching up
14:06:16 [AxelPolleres]
PROPOSED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2010-07-13
14:06:34 [LeeF]
look good to me
14:06:35 [AndyS]
seconded
14:06:36 [NickH]
UK number: "all circuits are busy now"
14:06:40 [AxelPolleres]
RESOLVED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2010-07-13
14:07:03 [AxelPolleres]
nick, can you scribe next week?
14:07:05 [Souri]
zakim, aaee is me
14:07:05 [Zakim]
+Souri; got it
14:07:38 [chimezie]
Axel: have new numbers, both should work but there are issues with UK numbers
14:07:46 [chimezie]
Sandro: may just be slow
14:07:49 [bglimm]
I can try dialing in again now
14:08:12 [chimezie]
AndyS: got an invalid number response
14:08:13 [bglimm]
I just had busy circuit
14:08:25 [Zakim]
-bglimm
14:08:40 [kasei]
regrets for next week, will be traveling
14:09:04 [chimezie]
Axel: need dedicated telecon for update model
14:09:14 [chimezie]
... Lee sent around link to doodle poll
14:09:17 [AxelPolleres]
doodle poll for update TC http://doodle.com/f83f6fe3e8g4gm5u
14:09:20 [Zakim]
+bglimm
14:09:30 [bglimm]
UK still does not work: busy circuits
14:09:56 [bglimm]
Zakim, mute me
14:09:56 [Zakim]
bglimm should now be muted
14:10:25 [kasei]
I'm interested in the TC, but if it's an issue of getting everyone together, I don't need to be there (as I listed quite a few days of unavailability)
14:11:34 [AxelPolleres]
let's fix Fri 30th July 4pm UK time, 11am Eastern time
14:11:47 [sandro]
(midnight in Japan. :-)
14:12:04 [AxelPolleres]
topic: approval of test cases
14:12:51 [LeeF]
Andy ran them successfully, I've looked at them and am happy with them.
14:13:02 [LeeF]
We accepted 1-4
14:13:05 [LeeF]
so 5-10
14:13:10 [Zakim]
+??P5
14:13:23 [chimezie]
AndyS: I execute them and get the right answers
14:13:29 [NickH]
Zakim, ??P5 is me
14:13:29 [Zakim]
+NickH; got it
14:13:40 [chimezie]
... tests have separate meanings now we moved things around
14:13:45 [chimezie]
Axel: Okay to approve?
14:13:47 [kasei]
+1
14:13:52 [AxelPolleres]
PROPOSED: approve subquery test cases 5-10
14:14:05 [chimezie]
Zakim, mute me
14:14:05 [Zakim]
Chimezie_Ogbuji should now be muted
14:14:06 [LeeF]
SteveH abstains
14:14:12 [LeeF]
(based on last week)
14:14:28 [AndyS]
who else has executed them?
14:14:38 [LeeF]
No one, as far as I know.
14:14:43 [LeeF]
No
14:14:54 [AndyS]
+1
14:15:36 [AxelPolleres]
RESOLVED: approve subquery test cases 5-10
14:16:08 [AndyS]
Thanks go to Olivier for creating the tests
14:16:19 [AxelPolleres]
issues on graph names in service description http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JulSep/0024.html
14:16:46 [chimezie]
Axel: there is some confusion WRT graph names when there are same as statements
14:17:00 [chimezie]
kasei: there is agreement that it is a non issue
14:17:09 [chimezie]
Sandro: issues are not about same as
14:17:16 [LeeF]
modelling issue
14:17:21 [chimezie]
kasei: disagreement on modeling WRT naming
14:17:35 [LeeF]
I think I agree with Greg - because of cases of things like having statistics about the graph in the context of a particular endpoint
14:17:48 [chimezie]
Sandro: my issue is that current model says there is a named graph, with a name, and the name of the graph is the graph itself
14:18:04 [chimezie]
kasei: wording in current query document, has that sentence
14:18:47 [chimezie]
Sandro: A URI has to be in quotes to be a name for a graph, pointy brackets are what they denote (<..> v.s. "..")
14:19:48 [chimezie]
... URI of graph and what it denotes is not the issue here. names are strings not things in domain of discourse
14:20:06 [AxelPolleres]
what'd be the issue with using xsd:anyURI literals as range?
14:20:39 [AndyS]
The named graph paper says the named graph is the pair (name, graph) not the graph.
14:20:47 [chimezie]
Sandro: no builtin in SPARQL to help with this
14:21:01 [kasei]
http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JulSep/0046.html
14:21:48 [AxelPolleres]
http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JulSep/0036.html
14:22:12 [sandro]
[] a sd:Dataset ;
14:22:12 [sandro]
sd:namedGraph <http://www.example/named-graph>.
14:23:38 [chimezie]
i'm not sure why using xsd:AnyUri isn't appropriate, seems to address the problem directly
14:24:24 [chimezie]
kasei: what about entailment? Sandro suggested that whenever a service had entailment on a named graph it should have a different name with the same underlying graph
14:24:43 [chimezie]
Sandro: entailment is a relationship between graphs that should have different URIs
14:24:58 [bglimm]
Zakim, unmute me
14:24:58 [Zakim]
bglimm should no longer be muted
14:25:05 [bglimm]
q+
14:25:06 [AxelPolleres]
q?
14:25:08 [kasei]
SELECT * FROM NAMED <a> WHERE { GRAPH <b> { ... } }
14:26:15 [AndyS]
Entailment is not a property of the graph deferred by <http://www.example/named-graph>
14:26:30 [chimezie]
bglimm: can't follow discussion, not sure what the problem is. Why need new name for <a>?
14:26:36 [LeeF]
I still don't think that entailment is the only case where the same graph might have different properties in the context of different endpoints
14:26:52 [chimezie]
Zakim, unmute me
14:26:53 [Zakim]
Chimezie_Ogbuji should no longer be muted
14:27:03 [AndyS]
q+
14:27:05 [LeeF]
access statistics, access control, ...
14:27:10 [ivan]
ack bglimm
14:27:13 [AxelPolleres]
q?
14:27:35 [AxelPolleres]
in which graph is <a> sameAs <b> ?
14:27:39 [chimezie]
I don't see the problem either
14:28:49 [chimezie]
... lots of conversation about clarifying entailment and graph identity ...
14:29:42 [AxelPolleres]
any's on the q ...
14:29:48 [bglimm]
ack me
14:30:17 [chimezie]
AndyS: 2 things. on SW when you deref URI you get data back involving sameAs assertions, how you treat them is a decision on the person pulling data in
14:30:25 [chimezie]
... not a property of the graph (a declarative structure)
14:30:48 [chimezie]
... there is no 'graph' with entailment
14:31:22 [bglimm]
for OWL there is not one graph with entailment, you just cannot build one graph that captures all entailments
14:31:29 [AxelPolleres]
SELECT * FROM NAMED <a> WHERE { GRAPH <b> { ... } }
14:31:37 [chimezie]
SELECT * FROM NAMED <a> WHERE { GRAPH <b> { ... } } -> [] , right?
14:33:02 [chimezie]
Sandro: RDF graph a contains RDFS statements (subClassOf statements)
14:33:06 [sandro]
<a> subclassof <b>. <b> subclass <c>.
14:33:18 [AxelPolleres]
<g1>
14:33:26 [sandro]
g1 is { <a> subclassof <b>. <b> subclass <c>. }
14:35:01 [chimezie]
Sandro: N3 has a different notion of graph. SPARQL notion of named graph is what i mean here
14:35:31 [AxelPolleres]
SELECT * FROM NAMED <g1> WHERE { GRAPH <g2> { ... } } -> [] ?
14:35:38 [sandro]
g2 is { <a> subclassof <b>. <b> subclassof <c>. <a> subclassof <c> }
14:36:01 [sandro]
g2 is { <a> subclassof <b>. <b> subclassof <c>. <a> subclassof <c>. ... }
14:36:02 [kasei]
infinite triples...
14:36:30 [ivan]
question, what is SELECT * FROM NAMED <g1> WHERE { GRAPH <g1> { ?x subclass ?y }} with RDFS semantics turned on?
14:36:40 [LeeF]
g2 is definitely not in the dataset there
14:36:59 [AxelPolleres]
the dataset is explicitly specified by FRPOM/FROM NAMED
14:37:01 [LeeF]
dataset definition in SPARQL is closed world :)
14:37:12 [Souri]
We do distinguish between a graph vs the entailed versions of that graph because a graph can be entailed with different entailment regimes. The way we do that is to allow an extra argument that specifies the entailment regime(s) being used.
14:37:23 [ivan]
q+
14:37:26 [LeeF]
Souri, do you give it a different URI?
14:37:26 [AndyS]
query is closed world ... it has to stop sometime :-)
14:37:31 [AxelPolleres]
q?
14:37:33 [AndyS]
q-
14:38:54 [LeeF]
ack ivan
14:39:37 [Souri]
No, we do not give it a different name, but we associate an "attachment" that includes entailment regime(s) and the combination of the query plus the attachment is sent over to Oracle DB.
14:39:42 [AndyS]
"""the relationship between an IRI and a graph in an RDF dataset is indirect."""
14:40:14 [chimezie]
There is conflation about what is in the graph (syntactically) and what is entailed (which is about what follows from the axioms, etc..)
14:40:49 [AndyS]
q+ to note there isn't always one entailed graph
14:40:52 [bglimm]
RDF Semantics also defines entailment between graphs and not entailment between a premise and a conclusion
14:41:00 [chimezie]
thinking about entailment as a relationship between two materialized graphs, is a misunderstanding of entailment
14:41:08 [bglimm]
we just don't do sub-graph matching any more
14:41:21 [ivan]
SELECT * FROM NAMED <g1> WHERE { GRAPH <g2> { ?x subclass ?y }; <g1> entails <g2> } is what one is considers...
14:41:24 [Souri]
We have been exploring a way to include the entailment regime(s) with standard graph names. So, for example FROM NAMED <g1> FROM NAMED orardf:rdfs ...
14:44:00 [Souri]
mega-graph == entailed graph?
14:44:15 [Souri]
q+
14:44:28 [AndyS]
q-
14:44:41 [bglimm]
mega-graph for logicians is a canonical model, but that might not help here ;-) entailed graph if there were only one entailed graph
14:44:46 [chimezie]
<a> subclassof <c> is not *in* (syntactically) <g1> but it is *entailed* by <g1> WRT RDFS
14:45:07 [AxelPolleres]
I'd rather say that the relationship between gragph and entailment regime is not direct...
14:45:25 [chimezie]
in and entailed are orthogonal relationships
14:46:44 [bglimm]
If we were strictly sticking to non-disjunctive formalisms, we could define everything in terms of deductive closures, but then OWL is out forever
14:46:51 [LeeF]
ack Souri
14:47:10 [chimezie]
Souri: have had difficulty in not specifying entailment regime
14:47:29 [ivan]
q+
14:47:34 [sandro]
YES. Changing the language to add an entailment parameter, as Souri says, would be much clearer
14:47:52 [AxelPolleres]
hmmm, this sounds like picking up parameterized entailment, but we ruled this out in the beginning
14:47:53 [chimezie]
... we thought of ways to guarantee to user that with certain arguments, results will match
14:47:58 [AndyS]
Jena currently requires different names for same graph (graph=value) under different entailment regimes.
14:48:04 [chimezie]
... will not be specified arbitrarily
14:48:19 [bglimm]
I would also like to be able to specify the regime in the query and not fixed per endpoint
14:48:26 [chimezie]
I agree, this is a big hole we have with entailment regime usage, that we have punted on
14:48:29 [bglimm]
but I thought it is out of scope
14:49:00 [chimezie]
Souri: we can use standard names to specify entailment regime within query
14:49:02 [sandro]
souri: we explored FROM NAME [some std name for OWL-entailment]
14:49:08 [LeeF]
Yes, it is something we discussed last year and did not make the scope of our work.
14:49:38 [LeeF]
http://www.w3.org/2009/sparql/wiki/Feature:ParameterizedInference
14:49:45 [AxelPolleres]
oracle does it with prefix... i.e. has some naming convention for inference.
14:50:03 [chimezie]
last, I checked the thread ended here: http://lists.w3.org/Archives/Public/public-rdf-dawg/2009AprJun/0051.html
14:50:48 [sandro]
q?
14:50:49 [LeeF]
ACTION: Lee to mark subquery tests 5-10 approved
14:50:49 [trackbot]
Created ACTION-280 - Mark subquery tests 5-10 approved [on Lee Feigenbaum - due 2010-07-27].
14:50:53 [chimezie]
Axel: we now have a definition of an entailment regime but not how to reference it explicitely
14:51:03 [chimezie]
... we are drifting away from original question
14:51:04 [AndyS]
Isn't this the role of service description now?
14:51:21 [sandro]
SELECT * FROM NAMED <g1> WHERE { GRAPH <g2> ENTAILS(e1) { ?x subclass ?y } }
14:51:26 [chimezie]
not sufficient, to allow the user to specify semantics of the query
14:51:33 [AndyS]
To do the *query* (algebra) under entailment, is not the same as BGPs under entailment.
14:51:52 [chimezie]
Axel: we are risking scope creep
14:51:54 [AxelPolleres]
q?
14:51:57 [LeeF]
ack ivan
14:52:02 [ivan]
SELECT * FROM NAMED <g1> WHERE { GRAPH <g2> { ?x subclass ?y }; <g1> entails <g2> }
14:52:06 [chimezie]
Ivan: respectfully disagree
14:52:17 [AndyS]
Curious: Why not SELECT * FROM NAMED <g2> { ... } ?
14:52:37 [chimezie]
... be explicit about forming graph via entailment and referring to it
14:52:53 [kasei]
AndyS, the whole point was my concern about impls that dereference g1 to construct the dataset
14:52:57 [bglimm]
I also wonder that AndyS
14:53:27 [AxelPolleres]
this looks like a new feature, which we haven't got on board....
14:53:36 [chimezie]
Ivan: it is a new SPARQl property, but the issue raised by Sandro can be addressed this way
14:53:57 [chimezie]
... either there in query or in service description
14:53:59 [AndyS]
kasei, sure - they get the (base) triples - impl can choose to understand the consequences or not.
14:54:18 [kasei]
q+
14:54:52 [chimezie]
kasei: putting it in SD won't always be enough for impl that dereference graphs to construct dataset
14:55:18 [chimezie]
Sandro: so don't support FROM NAMED entailment or add a keyword
14:56:11 [kasei]
I very much disagree with this "lying" notion
14:56:18 [LeeF]
Does anyone implement anything like this?
14:56:31 [LeeF]
Or, rather, who implements things like this?
14:56:45 [kasei]
LeeF, I deref URIs for datasets, and have experimental RDFS stuff that would run afoul of this
14:56:51 [bglimm]
As I understood Souri, Oracle is doing that
14:57:10 [chimezie]
Axel: need to follow up on email, perhaps
14:57:12 [AndyS]
"like" via different names in dataset but not via query syntax. Expected SD to describe the endpoint characteristics.
14:57:26 [bglimm]
Zakim, mute me
14:57:26 [Zakim]
bglimm should now be muted
14:57:54 [Souri]
Yes, we do allow user to specify entailment regime(s) in the query
14:58:15 [AxelPolleres]
ACTION: sandro to summarize the issue and mail his proposal to solve it
14:58:15 [trackbot]
Created ACTION-281 - Summarize the issue and mail his proposal to solve it [on Sandro Hawke - due 2010-07-27].
14:58:35 [sandro]
ACTION: sandro to take to e-mail the proposal, SELECT * FROM NAMED <g1> WHERE { GRAPH <g2> ENTAILS(e1) { ?x subclass ?y } }
14:58:35 [trackbot]
Created ACTION-282 - Take to e-mail the proposal, SELECT * FROM NAMED <g1> WHERE { GRAPH <g2> ENTAILS(e1) { ?x subclass ?y } } [on Sandro Hawke - due 2010-07-27].
14:58:43 [kasei]
briefly wanted to mention that I have a draft response to PA-2 at http://www.w3.org/2009/sparql/wiki/CommentResponse:PA-2
14:59:05 [chimezie]
Zakim, mute me
14:59:05 [Zakim]
Chimezie_Ogbuji should now be muted
14:59:14 [AxelPolleres]
ACTION: axel to mail a list of open issues and perceived status
14:59:14 [trackbot]
Created ACTION-283 - Mail a list of open issues and perceived status [on Axel Polleres - due 2010-07-27].
14:59:32 [Zakim]
-MattPerry
14:59:37 [bglimm]
bye
14:59:37 [Zakim]
-Souri
14:59:39 [Zakim]
-Chimezie_Ogbuji
14:59:40 [Zakim]
-Sandro
14:59:41 [NickH]
bye!
14:59:42 [Zakim]
-Lee_Feigenbaum
14:59:43 [AxelPolleres]
adjourned
14:59:43 [Zakim]
-bglimm
14:59:44 [Zakim]
-OlivierCorby
14:59:47 [Zakim]
-NickH
14:59:50 [Zakim]
-kasei
14:59:54 [Zakim]
-AndyS
14:59:55 [ivan]
zakim, drop me
14:59:55 [Zakim]
Ivan is being disconnected
14:59:56 [Zakim]
-Ivan
15:00:09 [Zakim]
-pgearon
15:00:16 [AxelPolleres]
rrsagent, make records public
15:01:02 [kasei]
current grammar doesn't allow selecing * along with other vars/expressions, right?
15:01:13 [AndyS]
right :-)
15:01:34 [kasei]
ok. maybe I'll fix the test cases that do that, then...
15:01:43 [AndyS]
Silly grammar.
15:01:44 [Zakim]
-AxelPolleres
15:01:45 [Zakim]
SW_(SPARQL)10:00AM has ended
15:01:46 [Zakim]
Attendees were +1.603.897.aaaa, Sandro, kasei, pgearon, +33.4.92.38.aabb, +1.603.897.aacc, Lee_Feigenbaum, MattPerry, AxelPolleres, OlivierCorby, Ivan, bglimm, AndyS,
15:01:48 [Zakim]
... +1.603.897.aaee, Chimezie_Ogbuji, Souri, NickH
15:01:59 [kasei]
silly? is it going to change?
15:02:27 [AndyS]
I proposed such a change. Steve is against it.
15:02:52 [kasei]
ah
15:03:42 [kasei]
probably shouldn't change anything just yet, then.
15:04:06 [AndyS]
I confess I don't see the problem as it's SQL rel alg feature commonly called "extend".
15:04:57 [AndyS]
rel alg "rename" does not (strictly) take an expression.
15:21:00 [SteveH_]
SteveH_ has joined #sparql
15:21:20 [SteveH_]
sorry, I wasn't able to dial in, some message about no lines free
15:32:06 [NickH]
SteveH_: I had the same problem and had to use US number again
15:41:14 [SteveH_]
I as on my mobile, so didn't want an intl call on the bill
16:03:47 [OlivierCorby]
OlivierCorby has left #sparql
17:03:00 [Zakim]
Zakim has left #sparql
17:48:56 [AxelPolleres]
AxelPolleres has left #sparql
18:10:10 [AndyS]
AndyS has joined #sparql
18:31:42 [AndyS]
AndyS has joined #sparql
20:26:25 [AndyS1]
AndyS1 has joined #sparql
21:45:13 [AxelPolleres]
AxelPolleres has joined #sparql
22:02:37 [SteveH_]
SteveH_ has joined #sparql
22:06:43 [sandro]
I'm thinking of posting a survey to semantic-web.
22:06:54 [sandro]
A quiz about Named Graphs.
22:07:16 [sandro]
(1) Is every Named Graph also a Graph T/F
22:09:10 [sandro]
(2) Named Graphs are sets of RDF triples (with an associated name) T/F
22:09:45 [sandro]
[er, (1) would be Every Named Graph is also a Graph ]
22:11:23 [sandro]
... and so on. Those are both False of course, and I expect most folks will get it wrong.
22:12:47 [sandro]
So, AxelPolleres, you really don't see a problem with popular usage of Graph being defn1 while the technically correct definition is defn2?
22:13:00 [sandro]
I mean:
22:14:01 [AxelPolleres]
well a named graph is a graph plus an identifier for it... so, it "is a graph" in a sloppy lingual sense, but not in a strict reading of "is" ?
22:14:23 [AxelPolleres]
so far for 1)
22:14:52 [AxelPolleres]
as for 2), I'd personally anwer no, because it is a det with an identifier
22:14:58 [AxelPolleres]
s/det/set/
22:15:14 [sandro]
Maybe: (1) conceptually, NamedGraph rdfs:subClassOf Graph (I'm not talking about sloppy linguistics)
22:15:35 [AxelPolleres]
hmm, not really.
22:15:43 [sandro]
Indeed, (1) is false.
22:16:04 [sandro]
I really hurts my C++ and Java trained brain to call it NamedGraph when it's not a subclass of Graph.
22:16:18 [AxelPolleres]
I haven't come across anyone getting it wrong yet or having a problem out of this distinction.
22:16:37 [AxelPolleres]
why? you have the same thing with objects in Java/C++ all the time...
22:16:56 [AxelPolleres]
two objects of type set are not the same, even if they have the same members.
22:17:07 [AxelPolleres]
?!?
22:17:50 [sandro]
If you name something [Adjective][NounPhrase] then by standard coding convetions, it must be a subclass of [NounPhrase].
22:19:12 [sandro]
GoodClient is a subclass of Client, etc.
22:19:50 [AxelPolleres]
Honestly, I think that the note in http://www.w3.org/2009/12/rdf-ws/Report.html just hits that point...
22:19:59 [AxelPolleres]
"(Note that the workshop participants realized that the term “named graph”, though widely used by the community, is ambiguous, and often refers to what could rather be referred to as quoted graphs, graph literals, etc. It was therefore decided to use the term “graph identification” for the purposes of reporting the workshop’s results, though this term is by no means definitive.)"
22:20:39 [AxelPolleres]
but I am pretty sure that we shouldn't bother about fixing this ambiguity, but rather RDF1.1 should.
22:22:58 [sandro]
Yeah -- of the ~10 points I've made today, that's one of the least important. Much more important is the KB vs Formula distinction. KB isn't exactly NamedGraph, it's close. One difference is that a KB can sensibly have multiple names.
22:24:02 [AxelPolleres]
depends on which setting you talk about.
22:24:44 [sandro]
This whole NamedGraphs vs KBs vs Formulas has to be properly formalized for metadata/provenance/etc to work right. Since we're not interoperable about that stuff yet, I guess it's not causing much pain.
22:25:19 [sandro]
which setting?
22:25:25 [AxelPolleres]
adn that's exactly what RDF1.1 needs to do, maybe.
22:25:58 [sandro]
Sure; I just don't want SPARQL saying anything more in the wrong direction.
22:26:19 [AxelPolleres]
when I talk about knolwdge bases in papers, I usually refer to a knoledge base by a name, but I didn't use/need/allow multiple names for a knowledge base.
22:26:39 [AxelPolleres]
"I just don't want SPARQL saying anything more in the wrong direction."
22:27:14 [AxelPolleres]
that's fine for me, then why not leave it as we have it now and not touch datasets in this round?
22:29:10 [sandro]
in toy/research use, I agree it doesn't make sense to have multiple names for kbs. But what's to stop me from saying a=kb1 in a suitable reasoning systems. Sure, you might not want to implement that, but it makes logical sense, demonstrating that the "naming" of KBs is EXACTLY the binding of URIs to objects in the domain of discourse -- not some other pairing function.
22:30:09 [AxelPolleres]
I don't get the distinction toy/research vs. suitable reasoning systems, sorry. ;-)
22:30:10 [sandro]
The current SD stuff has sparql formalizing NamedGraphs, and NamedGraphs as not being Graphs. That's saying more stuff in the wrong direction.
22:30:17 [sandro]
:-P
22:30:44 [sandro]
I mean, your systems so far are problem not concerned with reasoning about KBs by name. Right?
22:31:12 [AxelPolleres]
personally, I have no objections against naming graphs in SD with literals/strings instead of URIs
22:31:36 [sandro]
and calling them NamedGraphs ?
22:31:42 [AxelPolleres]
I do implement a system, for instance that implements the USING stuff you mentioned.
22:31:55 [sandro]
it's one thing to call them that in the editorial text of the spec; it's another to put that in running code.
22:34:14 [AxelPolleres]
(sarcasm alarm) we could even call them PinkElephants (serious again) we don;t have a better name, and I am worried about this discussion taking more time to end up with that same conclusion in the end
22:34:58 [sandro]
Yes. I was just going to say the most important thing you and I could be talking about is rif:import
22:35:29 [sandro]
but this SD stuff is somehow more compelling. *whacks head with book*
22:36:10 [AxelPolleres]
we just have a very tight charter and limited resources in what we can do in SPARQL1.1 for SPARQL2 we have to wait for what RDF1.1 does, I think, that's it.
22:36:25 [sandro]
http://www.w3.org/2005/rules/wiki/RIF_In_RDF#Importing_RIF_into_RDF
22:37:52 [AxelPolleres]
IIRC, I made a quite concrete proposal for formalising rif:imports in RDF to be added to SWC some longer time ago... this is something different, yes?
22:38:32 [sandro]
:-( I'm sorry, I don't remember the details. Any idea where that proposal would be? :-(
22:39:26 [sandro]
Mine is really just a sketch. I'd expect it to be the same as yours, but need more formal language. And my current name for it is rif:importedWithProfile which is of course arbitrary.
22:39:38 [sandro]
maybe rif:importedIntoNamedGraph would be better. :-) :-)
22:43:58 [AxelPolleres]
http://lists.w3.org/Archives/Public/public-rif-wg/2010Jan/0015.html
22:45:00 [AxelPolleres]
this was IMO simple, sweet and short, and the simplest among all the options listed in the more extensive mail I had sent earlier:
22:45:35 [AxelPolleres]
http://lists.w3.org/Archives/Public/public-rif-wg/2010Jan/0012.html (you might want to check for a list of all alternatives I had thought through)
22:46:26 [sandro]
Yeah, you're saying the same thing as me. rif:usedWithProfile owl:equivalentProperty rif:importedWithProfile
22:46:36 [AxelPolleres]
have to go now really, need to finish something for a deadline tonight and over that discussion didn't even start yet ;-)
22:46:45 [sandro]
Okay, enjoy!
22:46:49 [AxelPolleres]
cheers
22:46:53 [sandro]
errr, at least feel good about it. :-)
22:47:36 [AxelPolleres]
no worries, still have some enthusiasm left from vacation (only two days back)
22:47:47 [AxelPolleres]
AxelPolleres has left #sparql