From RDF Working Group Wiki
Revision as of 15:09, 26 September 2012 by Rcygania2
Please justify/explain non-obvious edits to this page, in your "edit summary" text.
14:56:03 <RRSAgent> RRSAgent has joined #rdf-wg 14:56:03 <RRSAgent> logging to http://www.w3.org/2012/09/19-rdf-wg-irc 14:56:05 <trackbot> RRSAgent, make logs world 14:56:05 <Zakim> Zakim has joined #rdf-wg 14:56:07 <trackbot> Zakim, this will be 73394 14:56:07 <Zakim> ok, trackbot; I see SW_RDFWG()11:00AM scheduled to start in 4 minutes 14:56:08 <trackbot> Meeting: RDF Working Group Teleconference 14:56:08 <trackbot> Date: 19 September 2012 14:56:24 <yvesr> Zakim, who is on the phone? 14:56:24 <Zakim> SW_RDFWG()11:00AM has not yet started, yvesr 14:56:25 <Zakim> On IRC I see RRSAgent, swh, Guus, AndyS, danbri, mischat1, MacTed, ivan, gkellogg, gavinc, manu, manu1, davidwood, yvesr, ericP, trackbot, sandro 14:58:29 <cygri> cygri has joined #rdf-wg 14:59:46 <cygri> zakim, code? 14:59:46 <Zakim> the conference code is 73394 (tel:+1.617.761.6200 sip:email@example.com), cygri 15:00:20 <Arnaud> Arnaud has joined #rdf-wg 15:00:27 <AndyS> zakim, start telecon 15:00:27 <Zakim> I don't understand 'start telecon', AndyS 15:00:37 <mlnt> mlnt has joined #rdf-wg 15:00:40 <davidwood> Zakim, I'm here... 15:00:40 <Zakim> I don't understand 'I'm here', davidwood 15:00:44 <AndyS> zakim, this is 73394 15:00:44 <Zakim> ok, AndyS; that matches SW_RDFWG()11:00AM 15:00:50 <yvesr> Zakim, who is on the phone? 15:00:56 <Zakim> On the phone I see ??P4, Guus, gavinc, ??P3, [IPcaller], davidwood 15:01:03 <yvesr> Zakim, ??P4 is me 15:01:04 <AndyS> zakim, IPcaller is me 15:01:08 <cygri> zakim, ??P3 is me 15:01:10 <Zakim> +Arnaud 15:01:10 <MacTed> trackbot, start call 15:01:10 <trackbot> Sorry, MacTed, I don't understand 'trackbot, start call'. Please refer to http://www.w3.org/2005/06/tracker/irc for help 15:01:18 <Zakim> +yvesr; got it 15:01:20 <Zakim> +AndyS; got it 15:01:28 <Zakim> +cygri; got it 15:01:32 <Zakim> +??P18 15:01:36 <gkellogg> zakim, ??P18 is me 15:01:52 <Zakim> +OpenLink_Software 15:01:57 <Zakim> +gkellogg; got it 15:02:02 <MacTed> Zakim, OpenLink_Software is temporarily me 15:02:04 <MacTed> Zakim, mute me 15:02:15 <MacTed> Zakim, who's here? 15:02:17 <Zakim> +MacTed; got it 15:02:21 <Zakim> MacTed should now be muted 15:02:33 <gavinc> ericP claimed he will be 15 minutes late 15:02:35 <Zakim> On the phone I see yvesr, Guus, gavinc, cygri, AndyS, davidwood, Arnaud, gkellogg, MacTed (muted) 15:02:52 <Zakim> On IRC I see mlnt, Arnaud, cygri, Zakim, RRSAgent, swh, Guus, AndyS, danbri, mischat1, MacTed, ivan, gkellogg, gavinc, manu, manu1, davidwood, yvesr, ericP, trackbot, sandro 15:03:04 <AlexHall> AlexHall has joined #rdf-wg 15:03:18 <Zakim> +Sandro 15:03:26 <ivan> zakim, dial ivan-voip 15:03:26 <Zakim> ok, ivan; the call is being made 15:03:28 <Zakim> +Ivan 15:03:42 <Zakim> +AlexHall 15:03:50 <ivan> zakim, mute me 15:03:50 <Zakim> Ivan should now be muted 15:04:01 <AndyS> can scribe for a bit 15:04:16 <Zakim> +??P24 15:04:23 <mlnt> zakim, ??P24 is me 15:04:23 <Zakim> +mlnt; got it 15:05:12 <AndyS> topic: admin 15:05:20 <AndyS> Minutes of 9/12 15:05:29 <davidwood> PROPOSED to accept the minutes of the 12 September: 15:05:29 <davidwood> http://www.w3.org/2011/rdf-wg/meeting/2012-09-12 15:05:43 <AndyS> topic; action items 15:05:49 <AndyS> david: none pending 15:06:11 <cygri> i claim victory on ACTION-183 15:06:14 <AndyS> RESOLVED accept the minutes of the 12 September: http://www.w3.org/2011/rdf-wg/meeting/2012-09-12 15:06:33 <AndyS> topic: action items 15:07:26 <sandro> action-183? 15:07:26 <trackbot> ACTION-183 -- Richard Cyganiak to send email to list about bNodes labels in TriG docs -- due 2012-09-12 -- OPEN 15:07:26 <trackbot> http://www.w3.org/2011/rdf-wg/track/actions/183 15:07:26 <davidwood> CLOSE ACTION-183 15:07:26 <trackbot> ACTION-183 Send email to list about bNodes labels in TriG docs closed 15:07:45 <AndyS> topic: next meeting 15:07:50 <AndyS> next wednesday 15:07:57 <AndyS> topic: F2F @ TPAC 15:08:21 <AndyS> davidwood: some people not said what state they are in 15:08:42 <AndyS> ... NB if attending, must register for TPAC as well. 15:09:02 <AZ> AZ has joined #rdf-wg 15:09:14 <sandro> https://www.w3.org/2002/09/wbs/35125/TPAC2012/registrants#RDF 15:10:23 <AndyS> davidwood: will follow up and sort out wiki page for attending F2F 15:10:35 <AndyS> sandro: some observer requests 15:10:38 <cygri> Fabien, EricP and Alexandre are registered but not on the wiki page 15:11:00 <Zakim> + +081165aaaa 15:11:23 <AZ> Zakim, +081165aaaa is me 15:11:23 <Zakim> +AZ; got it 15:11:36 <pfps> pfps has joined #rdf-wg 15:11:55 <MacTed> the wiki page for WG to update with TPAC/F2F attendance plans -- http://www.w3.org/2011/rdf-wg/wiki/FTF3 15:12:25 <AndyS> davidwood: only place is a limitation 15:12:40 <AndyS> sandro: probably room around the outside 15:13:53 <Zakim> +[GVoice] 15:13:55 <Zakim> +EricP 15:14:06 <gavinc> Zakim, mute ericP 15:14:06 <Zakim> EricP should now be muted 15:14:07 <pfps> zakim, GVoice is temporarily me 15:14:07 <Zakim> +pfps; got it 15:14:10 <AndyS> ericP arrives in a buzz 15:14:17 <Zakim> -EricP 15:14:19 <pfps> zakim, mute me 15:14:19 <Zakim> pfps should now be muted 15:14:20 <MacTed> Zakim, mute [GVoice] 15:14:20 <Zakim> sorry, MacTed, I do not know which phone connection belongs to [GVoice] 15:14:22 <Zakim> +EricP 15:14:41 <AndyS> scribenick: AndyS 15:14:45 <AndyS> scribenick: EricP 15:15:02 <cygri> http://www.w3.org/2011/rdf-wg/wiki/Meetings:Telecon2012.09.19 15:15:11 <sandro> sandro has changed the topic to: Meeting agenda, 2012-09-12: http://www.w3.org/2011/rdf-wg/wiki/Meetings:Telecon2012.09.19 15:15:16 <sandro> sandro has changed the topic to: Meeting agenda, 2012-09-19: http://www.w3.org/2011/rdf-wg/wiki/Meetings:Telecon2012.09.19 15:15:18 <Zakim> -cygri 15:15:34 <Zakim> +??P3 15:15:37 <cygri> zakim, ??P3 is me 15:15:37 <Zakim> +cygri; got it 15:16:02 <davidwood> Topic: Provenance Constraints Review 15:16:19 <davidwood> http://www.w3.org/blog/SW/2012/09/12/last-call-constraints-of-the-provenance-data-model/ 15:16:28 <ericP> davidwood: prov WG published a LC of their data model 15:16:31 <davidwood> Questions for the RDF WG: 15:16:31 <davidwood> - Does the terminology, Bundle and Document work with the terminology in the RDF WG? 15:16:31 <davidwood> - With respect to Bundle and Document do the defined constraints work with what is potentially being specified in RDF? 15:16:48 <davidwood> Zakim, who is on the phone? 15:16:48 <Zakim> On the phone I see yvesr, Guus, gavinc, AndyS, davidwood, Arnaud, gkellogg, MacTed (muted), Sandro, Ivan (muted), AlexHall, mlnt, AZ, pfps (muted), EricP, cygri 15:16:50 <ericP> ... they've asked for formal reivew, particularly of the above two questions 15:16:51 <cygri> i'll volunteer 15:18:43 <ericP> ACTION: cygri to review http://www.w3.org/blog/SW/2012/09/12/last-call-constraints-of-the-provenance-data-model/ by 2 Oct 15:18:43 <trackbot> Created ACTION-184 - Review http://www.w3.org/blog/SW/2012/09/12/last-call-constraints-of-the-provenance-data-model/ by 2 Oct [on Richard Cyganiak - due 2012-09-26]. 15:18:48 <ericP> ACTION: ericP to review http://www.w3.org/blog/SW/2012/09/12/last-call-constraints-of-the-provenance-data-model/ by 2 Oct 15:18:48 <trackbot> Created ACTION-185 - Review http://www.w3.org/blog/SW/2012/09/12/last-call-constraints-of-the-provenance-data-model/ by 2 Oct [on Eric Prud'hommeaux - due 2012-09-26]. 15:19:15 <davidwood> Topic: Turtle 15:19:24 <cygri> ACTION-184? 15:19:24 <trackbot> ACTION-184 -- Richard Cyganiak to review http://www.w3.org/blog/SW/2012/09/12/last-call-constraints-of-the-provenance-data-model/ by 2 Oct -- due 2012-09-26 -- OPEN 15:19:24 <trackbot> http://www.w3.org/2011/rdf-wg/track/actions/184 15:19:28 <Zakim> +Souri 15:19:38 <ericP> davidwood: LC for Turtle ended on the 15th 15:19:45 <Souri> Souri has joined #rdf-wg 15:19:54 <ericP> ... where are the comments being tracked? 15:20:15 <ericP> davidwood: we need to make sure we respond to every comment with a satisfactory answer 15:20:24 <ericP> ... we discussed using tracker or a wiki page 15:20:54 <ericP> ... either will give us a way to point to the comments 15:21:03 <pfps> We should try to respond to all LC comments, not just those in the window. 15:21:20 <cygri> see http://www.w3.org/2001/sw/rdb2rdf/wiki/Last_Call for another example 15:22:22 <ericP> ACTION: gavinc to add timbl's comment to tracker 15:22:22 <trackbot> Sorry, couldn't find user - gavinc 15:22:31 <ericP> ACTION: gavin to add timbl's comment to tracker 15:22:31 <trackbot> Created ACTION-186 - Add timbl's comment to tracker [on Gavin Carothers - due 2012-09-26]. 15:22:44 <ericP> close ACTION-186 15:22:44 <trackbot> ACTION-186 Add timbl's comment to tracker closed 15:23:11 <AndyS> IMO still need to respond to out-of-LC comments. 15:23:19 <ericP> gavinc: it's in tracker on the product "RDF Turtle" 15:23:37 <ericP> ... and yes, we need to respond to the out-of-LC comments: 15:23:53 <ericP> ... .. earlier this week: a way to remove prefix and base declarations 15:24:04 <ericP> ... .. a couple before LC, which i believe i've responded to 15:24:21 <ivan> -> http://lists.w3.org/Archives/Public/public-rdf-comments/2012Sep/0076.html the out-of-lc comment 15:24:25 <davidwood> Topic: JSON-LD 15:24:26 <ericP> ... i'll add those to tracker 15:24:54 <ericP> davidwood: LSON-LD is making good progress, but we need reviews from WG members who are not working on JSON-LD 15:25:40 <ericP> mlnt: we'd like reviews ASAP, of course, but it's not that urgent 15:25:50 <ericP> ... we're working on the last few issues 15:26:12 <ericP> ... these are about API so Syntax is stable enough for review 15:26:22 <cygri> yes sure 15:26:33 <mlnt> s/ LSON-LD/JSON-LD/ 15:26:45 <AlexHall> sorry, don't have the free time 15:26:45 <ivan> zakim, unmute me 15:26:45 <Zakim> Ivan should no longer be muted 15:28:14 <ivan> q+ 15:28:21 <davidwood> ack ivan 15:28:27 <ericP> gkellogg: iirc, there's some update to the preamble per cygri's review 15:28:38 <mlnt> Link to the latest JSON-LD Syntax Editor's Draft: http://dvcs.w3.org/hg/json-ld/raw-file/default/spec/ED/json-ld-syntax/20120830/index.html 15:28:44 <ericP> ivan: i'm reluctant to review when the TF isn't finished 15:29:00 <ericP> ... i'd rather here that they are done with syntax before starting reviews 15:29:19 <AndyS> +1 to Ivan - last reviews were on a changing doc which is less good for reviewers. 15:29:38 <MacTed> Zakim, who's noisy? 15:29:39 <ericP> mlnt: i think there's only one open issue -- checking 15:29:49 <Zakim> MacTed, listening for 10 seconds I heard sound from the following: davidwood (94%), mlnt (28%) 15:30:00 <MacTed> Zakim, mute mlnt 15:30:00 <Zakim> mlnt should now be muted 15:30:10 <mlnt> Open issues for syntax: https://github.com/json-ld/json-ld.org/issues?milestone=2&page=1&state=open 15:30:12 <cygri> i see 6 open syntax issues https://github.com/json-ld/json-ld.org/issues?labels=syntax&page=1&state=open 15:30:41 <mlnt> will mute myself MacTed 15:30:58 <Zakim> -cygri 15:31:14 <Zakim> +??P3 15:31:17 <cygri> zakim, ??P3 is me 15:31:17 <Zakim> +cygri; got it 15:31:38 <mlnt> Zakim, unmute me 15:31:38 <Zakim> mlnt should no longer be muted 15:31:43 <ericP> gavinc: there's an issue around alignment with RDF Concepts. is that tracked in tracker as well? 15:31:52 <gavinc> http://json-ld.org/spec/latest/json-ld-syntax/#relationship-to-other-linked-data-formats-and-data-models was confused by that issue marker 15:31:56 <ericP> gkellogg: i think that's 155 15:32:03 <mlnt> Explicit mapping of JSON-LD terminology to RDF terminology in Appendix "B1. RDF" -> assigned to cygri: https://github.com/json-ld/json-ld.org/issues/157 15:32:27 <Zakim> -pfps 15:32:41 <ericP> ACTION: gkellogg to notify cygri and sandro when the JSON-LD Syntax doc is ready for review 15:32:41 <trackbot> Created ACTION-187 - Notify cygri and sandro when the JSON-LD Syntax doc is ready for review [on Gregg Kellogg - due 2012-09-26]. 15:33:18 <ericP> topic: Graphs 15:33:45 <ericP> davidwood: there's a proposal from a few days ago that editors incorporate minimal semantics and tests 15:34:06 <davidwood> Zakim, who is on the phone? 15:34:06 <Zakim> On the phone I see yvesr, Guus, gavinc, AndyS, davidwood, Arnaud, gkellogg, MacTed (muted), Sandro, Ivan, AlexHall, mlnt, AZ, EricP, Souri, cygri 15:34:07 <ericP> ... since then, mailing list traffic indicates significant issues 15:34:18 <Zakim> +[GVoice] 15:34:28 <pfps> zakim, gvoice is temporarily me 15:34:28 <Zakim> +pfps; got it 15:34:58 <ericP> antoine: until PatH's last email, i thought we were making progress 15:35:35 <pfps> zakim, mute me 15:35:35 <Zakim> pfps should now be muted 15:35:35 <davidwood> Zakim, who is noisy? 15:35:38 <ericP> ... but still hoping for a minimal semantics, perhaps even more minimal than the current proposal, but some semantic requirements 15:35:46 <ivan> q+ 15:35:47 <Zakim> davidwood, listening for 10 seconds I heard sound from the following: davidwood (45%), AZ (5%) 15:36:12 <pfps> zakim, unmute me 15:36:12 <Zakim> pfps should no longer be muted 15:36:44 <ericP> pfps: my sentiments have been known for a long time 15:36:58 <ericP> ... i've been pointing out issues with the current proposal 15:37:44 <ericP> ... i think PatH was referring to a disagreement within the WG with what named graphs should mean, particularly in conjunction with the default graph 15:38:15 <ericP> davidwood: do you believe we can define a more minimal dataset semantics which could be standardized? 15:38:41 <ericP> pfps: i originally proposed that the only semantics should be for graphs. 15:38:42 <ivan> q- 15:38:48 <AZ> q+ 15:39:26 <ericP> ... certain things about the default graph make big changes the meaning of the named graphs 15:39:50 <ericP> ... e.g. inconsistent default graph makes the named graphs irrelevent 15:40:01 <ivan> -> http://lists.w3.org/Archives/Public/public-rdf-wg/2012Sep/0180.html Pat's email 15:40:16 <ericP> ... this is frighteningly powerful 15:40:43 <ivan> "Second, I do not see why we need to give a semantics for datasets. " from Pat's email 15:40:48 <ericP> ... contigent equalities in the default graph could imply contigent equalities in the name graphs which could necessitate merges 15:41:03 <ericP> ... could imply high reasoning requirements 15:41:08 <MacTed> Zakim, unmute me 15:41:08 <Zakim> MacTed should no longer be muted 15:41:10 <ivan> After reading and following all this debate, I agree with Peter. 15:42:05 <ivan> (^ was a quote from Pat's mail) 15:42:17 <ericP> ... you can define entailment for RDF datasets without defining interpreation for datasets 15:42:55 <pfps> dataset entailment can be defined in terms of entailment between the constituent graphs (several ways) 15:43:05 <gkellogg> s/interpreation/interpretation/ 15:43:10 <PatH> PatH has joined #rdf-wg 15:43:11 <ericP> davidwood: given that, i'm starting to understand this debate 15:43:19 <davidwood> q? 15:43:21 <davidwood> ack AZ 15:43:46 <Zakim> +PatH 15:43:47 <ericP> AZ: do you have a concrete example that, even if the default graph is inconsistent, that the whole dataset is inconsistent? 15:44:00 <PatH> sorry im late 15:44:38 <ericP> pfps: if the default graph is the merge of the named graphs, and you're merging to be able to look at everything, you wouldn't want an inconsistent merge render the named graphs inconsistent or trivial 15:44:42 <ericP> q? 15:44:46 <MacTed> q+ 15:45:15 <PatH> +1 to what peter just said. 15:45:22 <ericP> pfps: you wouldn't want inconsistencies in the default graph make the whole dataset inconsistent? 15:45:39 <cygri> perhaps globally rename "dataset interpretation" as "default graph interpretation"? 15:45:46 <ericP> q+ to propose an OWL inconsistency and ask what it does to a dataset 15:46:13 <ericP> pfps: people could ingore the minimal or stronger semantics and proceed as usual, but they would be out of standard 15:46:25 <PatH> no property of any part of the dataset should mean that the dataset as a whole has any property. 15:46:36 <Zakim> -cygri 15:46:55 <Zakim> +??P3 15:46:55 <ericP> ... would then imply that SPARQL queries on named graphs may be ... inconsistent? 15:46:57 <PatH> q+ 15:47:07 <cygri> zakim, ??P3 is me 15:47:07 <Zakim> +cygri; got it 15:47:36 <davidwood> ack MacTed 15:47:42 <ericP> AZ: iirc, if you try to query an graph inconsistent with respect to the entailment regime, the system has to return an error 15:47:45 <pfps> zakim, mute me 15:47:45 <Zakim> pfps should now be muted 15:47:49 <MacTed> Zakim, unmute me 15:47:49 <Zakim> MacTed was not muted, MacTed 15:48:15 <ericP> MacTed: is the default graph not an implementation-dependent detail? 15:48:38 <ericP> ... sometimes the union of all graphs, sometimes its own thing 15:49:04 <ericP> ... that the union of all named graphs is inconsistent is de rigor 15:49:10 <pfps> the default graph can be anything - the merge, metadata, something unrelated, .... 15:49:28 <PatH> is there a requirement that the default graph be the union of named graphs? 15:49:29 <ericP> ... this sounds like we're going back to saying that anything in RDF must be real truth forevermore 15:49:48 <ericP> davidwood: i've never seen a real-world consistent dataset 15:49:52 <MacTed> s/de rigor /de rigeur/ 15:50:14 <gavinc> no. 15:50:25 <ericP> PatH: i think we have to understand "inconsistency" 15:50:38 <ericP> ... it's hard to create inconsistency just in RDF 15:50:51 <ericP> davidwood: but folks say "it will break OWL" 15:51:10 <davidwood> ack ericP 15:51:10 <Zakim> ericP, you wanted to propose an OWL inconsistency and ask what it does to a dataset 15:51:42 <gkellogg> lastModified duplication inconsistency too. 15:51:47 <AlexHall> is it possible for a graph to be RDF-Simple inconsistent? 15:52:19 <cygri> AlexHall, yes, via broken rdf:XMLLiterals 15:52:28 <PatH> alexhall, yes but only by doing something insane with an xml literal 15:52:34 <AndyS> Alex - with RDF XML lIterals (only, IIRC) Only datatype fixed in RDF-basic 15:52:39 <AlexHall> ahh, forgot about that 15:52:46 <AZ> In SPARQL 1.1 Entailment Regimes: The scoping graph is graph-equivalent to the active graph even if the active graph is RDFS-inconsistent. If the active graph is RDFS-inconsistent, an implementation MAY raise a QueryRequestRefused fault or issue a warning and it SHOULD generate such a fault or warning if, in the course of processing, it determines that the data or query is not compatible with the request. In the presence of an inconsistency the conditions on solut 15:53:11 <AZ> (this is for RDFS but it's essentially the same for other regimes) 15:53:42 <AlexHall> does that still hold with the XMLLiteral changes in 1.1? 15:54:27 <AlexHall> aren't they optional now, i.e. not folded into basic RDF? 15:54:30 <PatH> alex, likely not. 15:54:40 <cygri> AlexHall, yes. XMLLiterals still have to be well-formed XML. the change was only that they no longer need to be normalized 15:55:00 <AlexHall> got it, thanks. 15:55:06 <AZ> cygri, with "simple entailment", all RDF graphs are "simple-consistent" 15:55:29 <ericP> sandro: agreeing with ericP [he can do that?], it would be nice to have a tree of reference 15:55:42 <AZ> inconsistency only occurs in RDFS and its extensions 15:55:49 <cygri> AZ, AlexHall: right, simple entailment != RDF entailment. i was talking about RDF entailment 15:55:50 <ericP> ... i view the default graph as the place to put the metadata because i don't know where else to put it 15:57:00 <AlexHall> ah yes, "Simple Entailment between RDF graphs" != "RDF Entailment" 15:57:20 <AndyS> Common case - one graph - no named graphs. 15:58:45 <cygri> sandro++ 16:00:35 <davidwood> ack PatH 16:00:38 <davidwood> q? 16:00:46 <ericP> ... folks seem pretty comfortable with that notion but it bumps up into the occasional SPARQL usage of the default graph as a union of all graphs 16:01:08 <ericP> ericP: or the common SPARQL usage with a pre-loaded default graph 16:01:23 <ericP> PatH: there are several ways default graphs are being used 16:01:44 <ericP> ... so we don't want to impose a relationship between default graph and named graphs 16:01:47 <ivan> q+ 16:01:59 <AZ> q+ 16:02:02 <pfps> what is the relationship that SPARQL suggests? 16:02:16 <Zakim> -cygri 16:02:35 <davidwood> ack ivan 16:02:44 <ericP> ... so maybe we should should not define that at all 16:02:53 <ericP> ivan: PatH was quicker than i 16:03:01 <Zakim> +??P13 16:03:05 <cygri> zakim, ??P13 is me 16:03:05 <Zakim> +cygri; got it 16:03:17 <ericP> ... are we facing such divergence that it's wisest not to standardize dataset semantics 16:03:29 <ericP> ... that said, it could be nice to document what we have 16:03:58 <sandro> q+ 16:04:09 <ericP> ... e.g. publish the wiki as a note documenting one approach 16:04:38 <cygri> the main troublesome divergent practice is "default graph for metadata" vs "default graph as union graph" 16:04:43 <davidwood> ack AZ 16:04:56 <sandro> q+ to say okay with no semantics for datasets, but trig still needs some way to include metadata. 16:05:01 <ericP> davidwood: we need to document what we have, but not sure it serves much of the community 16:05:18 <Zakim> -gkellogg 16:05:18 <ericP> AZ: there shouldn't be a relationship between the default graph and the named graphs 16:05:30 <PatH> cygri, there also seems to be default as a distillation of the consistent parts of the named graphs. 16:05:50 <cygri> PatH, not sure what that means. where have you seen this? 16:05:53 <pfps> zakim, mute me 16:05:53 <Zakim> pfps was already muted, pfps 16:06:04 <ericP> ... i propose something weaker, that inconsistency in the default graph means the name graphs are inconsistent 16:06:14 <ivan> zakim, who is noisy? 16:06:14 <davidwood> PatH, that would be news to me (and horrible to implement) 16:06:23 <davidwood> q? 16:06:24 <Zakim> ivan, listening for 10 seconds I heard sound from the following: gavinc (55%), MacTed (5%) 16:06:29 <Zakim> +??P73 16:06:33 <davidwood> ack sandro 16:06:33 <Zakim> sandro, you wanted to say okay with no semantics for datasets, but trig still needs some way to include metadata. 16:06:35 <ericP> ... i think the union default graph is mostly internal, in which case it's not *the* default graph 16:06:38 <gkellogg> zakim, ??P73 is me 16:06:43 <Zakim> +gkellogg; got it 16:06:55 <ericP> sandro: i think that a lot of the issues are coming from pfps's issues with the default graph 16:07:12 <ericP> ... noting ivan's suggestion that we skip semantics 16:07:17 <PatH> cygri, isnt this what people are talking about when they talk about the need to keep dirty data clearly separated intro the named graphs? 16:07:27 <ericP> ... SPARQL 1.1 Service Description has a place to store metadata 16:07:47 <cygri> PatH, no, that doesn't involve the default graph at all 16:07:47 <gavinc> ... what huh? Why the heck do I need a magic "metadata" graph? 16:08:11 <PatH> cygri, then i am completely confused. 16:08:14 <ericP> ... there are other ways we could store it in trig 16:08:15 <AndyS> so document use cases and solutions? 16:08:40 <ericP> davidwood: why do we have disagreement on this simple use case 16:09:04 <PatH> q+ to ask dumb clarification question 16:09:20 <ericP> ... sandro wants to grab a gtext from the web, shove it in a gbox, and talk about it 16:09:29 <pfps> I have nothing against using the default graph to store metadata. I'm just against requiring the default graph for that purpose. It seems to me that there are very many potential uses for the default graph, and none of them dominate. 16:09:31 <cygri> PatH, that's why we need *named graphs* in the first place. it doesn't say why we need a default graph 16:09:36 <ericP> ... this sue case still seems to be causing dissention 16:09:46 <gavinc> I think this is a fine use case, I just do NOT understand why there needs to be MORE MAGIC 16:09:52 <davidwood> ack PatH 16:09:52 <Zakim> PatH, you wanted to ask dumb clarification question 16:09:53 <ericP> ... why wouldn't we want a dataset in order to describe it? 16:09:55 <pfps> +1 to gavin 16:10:43 <ericP> PatH: re: the irc discussion above, since day 1 we've noted that data on the web is dirty so it needs to be stored in separate named graphs 16:10:59 <ericP> ... i've been assuming that default graph was the place where you stored what you believed 16:11:16 <pfps> I don't think that Pat is wrong, he is just thinking of a particular use of the default graph. 16:11:31 <MacTed> Zakim, unmute me 16:11:31 <Zakim> MacTed was not muted, MacTed 16:11:34 <ericP> ... if the reason for having named graphs is to compartmentalize dirt, what's the purpose of the default graph 16:12:08 <sandro> pat: I thought the idea was to put the accepted part of the data into the default graph 16:12:12 <ericP> cygri: you mentioned "putting the accepted parts of the named graphs in the default graph" 16:12:18 <sandro> cygri: well, the metadata, at least. 16:12:28 <pfps> Huh, metadata isn't data (or at least certainly isn't *all* data, or even all clean data). 16:12:48 <ericP> ... i've seen people store metadata in the named graphs, but i've not seen certain named graphs annointed and imported into the default graph 16:13:00 <ivan> q+ 16:13:24 <ericP> PatH: so i should model the typical use of the default graph as being to hold metadata 16:13:27 <gavinc> But you can put the metadata in ANOTHER named graph 16:13:31 <sandro> q+ to say it's about accountability (provenance, feedback) not dirt, per se. 16:13:47 <ericP> ... so when i'm grabbing diverse data, is should think of the default graph as metadata? 16:14:23 <davidwood> Ted: There are no underlying truths of the (Web) universe 16:14:31 <ericP> MacTed: in the scenario you described, the default graph should only hold underlying truths of the universe 16:14:46 <sandro> q? 16:14:57 <pfps> one can also think of using the default graph for the main data, and named graphs for other purposes (e.g., working with annotations) - Bijan proposed this in the OWL WG (not using datasets of course) 16:15:00 <sandro> davidwood, extend meeting? this eems kinda productive. 16:15:02 <ericP> ... i agree with sandro, [<sandro> q+ to say it's about accountability (provenance, feedback) not dirt, per se.] 16:15:07 <davidwood> sandro, yes 16:15:13 <gavinc> The default graph is the graph that is queried when you don't say use a specific graph :P 16:15:18 <ericP> ... the default graph is an implementation detail, not an RDF detail 16:15:21 <gavinc> The "default" 16:15:23 <sandro> +1 extend 16:15:29 <davidwood> q? 16:15:33 <davidwood> ack ivan 16:15:40 <davidwood> Thanks, pfps 16:15:44 <Zakim> -pfps 16:16:20 <ericP> ivan: per what MacTed said, some SPARQL engines use a union of named graphs as the default graph 16:16:26 <sandro> ivan: We may be diverted by SPARQL's union-default graphs. That's just an implementation thing, not something we should worry about. 16:16:28 <ericP> ... we shouldn't take that into account 16:16:32 <cygri> ivan++ 16:16:44 <AZ> +1 ivan 16:16:45 <ericP> ... what we model is a set of graphs, one called the default 16:16:47 <gavinc> Default exists at QUERY time. 16:16:51 <Souri> I agree with Ivan: unnamed graph != SPARQL default graph 16:16:52 <sandro> +1 ivan (let's not worry about union-default graphs) 16:16:58 <ericP> q+ to talk about *fixed* default graphs 16:16:58 <gavinc> It doesn't "exist" strongly 16:17:18 <gavinc> "a default graph consisting of the RDF merge of the graphs referred to in the FROM clauses, and" 16:17:25 <PatH> ok, thanks for the explanation. I conclude that it is even more important for us to to NOT impose ANY semantic relationship between the default graph and the dataset. 16:17:31 <davidwood> ack sandro 16:17:31 <Zakim> sandro, you wanted to say it's about accountability (provenance, feedback) not dirt, per se. 16:17:51 <ericP> davidwood: we got into this in order to address SPARQL, but maybe it's ok if it's not a core RDF point 16:17:51 <gavinc> +q the ONLY defined form of "default" graph from SPARQL query IS as a merge 16:17:56 <sandro> PROPOSED: RDF-WG is not concerned about compatibility with some SPRARLQ implemtnation union-of-graphs default graph? 16:17:58 <gavinc> +q to say the ONLY defined form of "default" graph from SPARQL query IS as a merge 16:18:14 <Zakim> -cygri 16:18:20 <AZ> I like this proposition 16:18:27 <Zakim> +??P13 16:18:43 <cygri> zakim, ??P13 is me 16:18:43 <Zakim> +cygri; got it 16:18:48 <ericP> sandro: re: PatH's discussion of dirty data in named graphs, it's more about provenance than dirt 16:19:13 <PatH> OK, I was speaking informally, obviously. 16:19:14 <ericP> ... folks downstream can make decisions about the utility/cleanliness 16:19:25 <davidwood> q? 16:19:31 <davidwood> ack ericP 16:19:31 <Zakim> ericP, you wanted to talk about *fixed* default graphs 16:21:09 <PatH> surely the default graph is "special" in some way. 16:21:12 <davidwood> ack gavinc 16:21:12 <Zakim> gavinc, you wanted to say the ONLY defined form of "default" graph from SPARQL query IS as a merge 16:21:44 <sandro> PROPOSED: The RDF-WG in considering semantics of Datasets is not guided by concerns of compatibility with those SPARQL engines which choose to make their default graph automatically be the union of all the named graphs. 16:21:45 <davidwood> http://www.w3.org/TR/sparql11-query/#rdfDataset 16:22:06 <cygri> q+ 16:22:07 <ericP> ericP: many SPARQL implementations have a prescribed default graph. 16:22:30 <AZ> Should we distinguish "SPARQL dataset" and "RDF dataset"? 16:22:53 <ericP> ... for their use cases, it must have certain contents. it may be uncomfortable for someone else to say that it has to contain other stuff 16:22:54 <cygri> most SPARQL queries don't use FROM. 16:22:54 <PatH> from the above, 16:22:57 <PatH> " 16:23:02 <AndyS> Most queries do not have FROM in them. Forget that part for this discussion. There is a dft graph to query - it just "is". 16:23:05 <PatH> "13.1 Examples of RDF Datasets The definition of RDF Dataset does not restrict the relationships of named and default graphs." 16:23:20 <PatH> +1 to last speaker 16:24:01 <cygri> FROM != FROM NAMED 16:24:03 <ericP> gavinc: we're treating the default graph as fixed, but SPARQL queries with FROM change the default graph 16:25:02 <PatH> what??? 16:25:07 <davidwood> The default graph formed by a SPARQL query is *transient* 16:25:09 <sandro> PROPOSED: The RDF-WG in considering semantics of Datasets is not guided by concerns of compatibility with those SPARQL engines which choose to make their default graph automatically be the union of all the named graphs. 16:25:18 <ivan> +1 16:25:29 <gavinc> -0.9 16:25:33 <gkellogg> +1 16:25:37 <sandro> +0.5 16:25:42 <Souri> +1 16:25:44 <PatH> -1 16:25:50 <AZ> +1 16:25:57 <davidwood> +1 16:26:02 <PatH> we should be **compatible** 16:26:02 <cygri> gavinc, FROM and FROM NAMED both build a new dataset. most SPARQL queries use a pre-existing dataset that is provided by the store/service. 16:26:06 <AndyS> 0 (OK-ish but bad framing for a significant minority - the "context" people) 16:26:30 <sandro> +1 <cygri> gavinc, FROM and FROM NAMED both build a new dataset. most SPARQL queries use a pre-existing dataset that is provided by the store/service. 16:26:33 <AndyS> +1 to cygri 16:26:41 <davidwood> PatH, we should be compatible, yes. But there may be no compatibility issue here. 16:26:57 <ericP> i think "union" is a red herring. what we care about is those SPARQL endpoints with a prescribed default graph 16:26:57 <Arnaud> 0 (would rather know how common this is before deciding) 16:27:21 <sandro> great meeting, guys. :-) 16:27:31 <PatH> david, but as stated it says we should not consider cvompatibility. bad wording. 16:27:49 <AndyS> There is a dataset (G, (ni,Gi)), it is queried. 16:27:53 <AlexHall> default graph as union is an implementation detail. it can be treated as a distinct graph at query time and not run afoul of anything we're doing here. 16:28:26 <AndyS> +1 to Alex 16:28:30 <gavinc> +1 16:28:39 <PatH> but that is irrelevant. 16:28:52 <Zakim> -gkellogg 16:28:53 <Zakim> -Souri 16:28:54 <Zakim> -Ivan 16:28:55 <Zakim> -AlexHall 16:28:55 <Zakim> -MacTed 16:28:56 <Zakim> -cygri 16:28:56 <Zakim> -PatH 16:28:58 <Zakim> -mlnt 16:28:59 <Zakim> -EricP 16:29:05 <Zakim> -davidwood 16:29:07 <Zakim> -Arnaud 16:29:09 <Zakim> -Sandro 16:29:11 <Zakim> -gavinc 16:29:13 <Zakim> -AndyS 16:29:17 <Zakim> -yvesr 16:29:20 <AlexHall> AlexHall has left #rdf-wg 16:29:29 <Zakim> -AZ 16:30:26 <davidwood> Quoting from http://www.w3.org/TR/sparql11-query/#rdfDataset "A SPARQL query can match different parts of the query pattern against different graphs as described in section 13.3 Querying the Dataset." 16:30:55 <davidwood> That section (13.3) starts by saying, "When querying a collection of graphs" 16:31:30 <davidwood> Therefore, I'm not sure RDF datasets as defined in SPARQL apply when FROM and FROM NAMED aren't used. 16:31:43 <AndyS> 13.3 : A SPARQL query *may* specify the dataset to be used for matching by using the FROM clause and the FROM NAMED clause to describe the RDF dataset. 16:31:45 <davidwood> However, I'll take the advice and read more. 16:32:11 <Zakim> -Guus 16:32:14 <Zakim> SW_RDFWG()11:00AM has ended 16:32:15 <Zakim> Attendees were Guus, gavinc, davidwood, Arnaud, yvesr, AndyS, cygri, gkellogg, MacTed, Sandro, Ivan, AlexHall, mlnt, AZ, EricP, pfps, Souri, PatH 16:32:20 <AndyS> "If a query provides such a dataset description ..." 16:32:25 <Zakim> apparently SW_RDFWG()11:00AM has ended, tlr 16:32:29 <Zakim> On IRC I see tlr, PatH, pfps, Arnaud, Zakim, RRSAgent, Guus, AndyS, mischat1, MacTed, gkellogg, gavinc, manu, manu1, davidwood, yvesr, ericP, trackbot, sandro 16:33:00 <davidwood> AndyS, I think you mean section 13.2 16:33:36 <AndyS> probably. 16:33:36 <davidwood> "A SPARQL query may specify the dataset to be used for matching by using the FROM clause and the FROM NAMED clause to describe the RDF dataset. If a query provides such a dataset description, then it is used in place of any dataset that the query service would use if no dataset description is provided in a query." 16:33:42 <tlr> tlr has left #rdf-wg 16:34:03 <davidwood> Right, so there is a notion of a "default dataset", which is completely (as far as I can tell) implementation dependent. 16:34:40 <AndyS> All data at an end point is impl dependent. (That's why you go there!:-)) 16:35:00 <davidwood> That is, not defined in any way by the RDF family of specifications, nor in SPARQL specs. It is up to implementations of a SPARQL endpoint how to define such a default dataset. Is that right? 16:36:19 <AndyS> A service offers (1) some dataset or (2) a way to build one. (1) is most common on the web by a long way. There is also examples of "pick from local storage". 16:37:38 <davidwood> Yes, I think we agree now. 16:37:40 <davidwood> Thanks 16:37:42 <AndyS> "union" is just a way to explain where dft came from (esp. "context" people). Not relevant at query time. 16:38:05 <davidwood> Sorry it took me so long to page in the spec. 16:39:48 <AndyS> Few web endpoints offer general load-from-web-and-query for some strange reason. 16:43:18 <davidwood> Security (DoS attacks). Any such system could be brought to its knees too easily. 16:43:59 <davidwood> I'm tempted to bring up a service like that on EC2, though, with throwaway machines that kill themselves on paging. 16:48:18 <danbri> danbri has joined #rdf-wg 17:07:02 <MacTed> URIBurner.com does a load-from-web (queued task) and query (all data is available through SPARQL endpoint once the load's done)... of course, there are various throttles in place 17:26:09 <cygri> cygri has joined #rdf-wg 17:39:21 <danbri> danbri has joined #rdf-wg 17:53:49 <davidwood> Yeah, I use URIBurner every now and then. It is a cool and useful service. 18:15:49 <danbri> danbri has joined #rdf-wg 18:20:27 <danbri> danbri has joined #rdf-wg 18:57:26 <mischat> mischat has joined #rdf-wg 18:58:14 <Zakim> Zakim has left #rdf-wg 19:41:17 <cygri> cygri has joined #rdf-wg 19:56:34 <gkellogg> gkellogg has joined #rdf-wg 20:34:26 <danbri> danbri has joined #rdf-wg 20:34:51 <danbri_> danbri_ has joined #rdf-wg 21:35:24 <Arnaud> Arnaud has joined #rdf-wg 21:35:35 <Arnaud> Arnaud has left #rdf-wg 22:08:46 <danbri> danbri has joined #rdf-wg # SPECIAL MARKER FOR CHATSYNC. DO NOT EDIT THIS LINE OR BELOW. SRCLINESUSED=00000518