Warning:
This wiki has been archived and is now read-only.
Chatlog 2011-12-21
From RDF Working Group Wiki
See panel, original RRSAgent log or preview nicely formatted version.
Please justify/explain non-obvious edits to this page, in your "edit summary" text.
16:02:08 <trackbot> Meeting: RDF Working Group Teleconference 16:02:08 <trackbot> Date: 21 December 2011 16:02:13 <AndyS> zakim, who is on the phone? 16:02:13 <Zakim> SW_RDFWG()11:00AM has not yet started, AndyS 16:02:14 <Zakim> On IRC I see Scott_Bauer, cygri, pfps, AndyS, AZ, pchampin, JeremyCarroll, Zakim, RRSAgent, LeeF, gavinc, Guus, MacTed, mischat, manu, trackbot, davidwood, mdmdm, manu1, NickH, 16:02:15 <gavinc> zakim, this is 73394 16:02:17 <Zakim> ... sandro, ericP 16:02:17 <Zakim> ok, gavinc; that matches SW_RDFWG()11:00AM 16:02:41 <gavinc> there we go! 16:02:51 <Guus> zakim, who is here? 16:02:51 <Zakim> On the phone I see +31.20.598.aaaa, Sandro, Peter_Patel-Schneider, +1.540.898.aabb, [IPcaller], +1.707.861.aacc, +1.415.586.aadd, ??P9, ??P11 16:02:53 <Zakim> On IRC I see Scott_Bauer, cygri, pfps, AndyS, AZ, pchampin, JeremyCarroll, Zakim, RRSAgent, LeeF, gavinc, Guus, MacTed, mischat, manu, trackbot, davidwood, mdmdm, manu1, NickH, 16:02:55 <Zakim> ... sandro, ericP 16:02:56 <AndyS> scribe: AndyS 16:03:28 <Zakim> On the phone I see Guus, Sandro, Peter_Patel-Schneider, +1.540.898.aabb, AndyS, gavinc, JeremyCarroll, ??P9, ??P11, MacTed (muted), cygri 16:03:37 <AndyS> scribenick: AndyS 16:04:17 <AndyS> topic: Admin 16:05:42 <AndyS> david: admin section then discussion of the path forward for [graph] 16:05:46 <davidwood> PROPOSED to accept the minutes of the 14 Dec telecon: 16:05:46 <davidwood> http://www.w3.org/2011/rdf-wg/meeting/2011-12-14 16:05:51 <Guus> zakim, who is on the phone? 16:05:51 <Zakim> On the phone I see Guus, Sandro, Peter_Patel-Schneider, +1.540.898.aabb, AndyS, gavinc (muted), JeremyCarroll (muted), yvesr, AZ, MacTed (muted), cygri, Scott_Bauer, AlexHall 16:06:10 <davidwood> zakim, aabb is me 16:06:10 <Zakim> +davidwood; got it 16:06:28 <AndyS> topic: Action item review 16:06:35 <AndyS> No pending action items 16:06:45 <AndyS> David: 13 open actions 16:07:06 <Zakim> + +1.760.705.aaff 16:07:07 <danbri> danbri has joined #rdf-wg 16:07:13 <pchampin> zakim, aaff is me 16:07:13 <Zakim> +pchampin; got it 16:07:16 <pchampin> zakim, mute me 16:07:16 <Zakim> pchampin should now be muted 16:08:15 <gavinc> Zakim, unmute me 16:08:15 <Zakim> gavinc should no longer be muted 16:09:07 <AndyS> david: action 117 - jeremy ... 16:09:13 <AndyS> ... duration datatypes 16:09:27 <AndyS> jeremy: will do action 16:09:37 <AndyS> ... action 118 similar 16:10:09 <AndyS> david: gavin - action on formatted text literals 16:10:14 <cygri> q+ 16:10:55 <AndyS> ack cygri 16:11:08 <AndyS> cygri: pls expand on action 124 16:11:28 <AndyS> gavin: it's because of xml literal downplayed ... what about formatted text? 16:11:55 <AndyS> topic: next meeting 16:12:01 <AndyS> Jan 4 16:12:07 <AndyS> topic: path forward 16:12:45 <AndyS> david: Named graphs has been a major matter for the WG ... but also now see there are concerns about how people view the web 16:13:09 <AndyS> ... lot of perspectives in WG and these all be considered valid. 16:13:34 <AndyS> ... new uses of RDF (Linked Data Ent Workshop) 16:13:50 <AndyS> ... not just for one thing 16:14:09 <AndyS> ... pls keep in mind that other people's perspective are valid 16:14:42 <AndyS> ... WG needs to find a compromiose 16:15:00 <AndyS> ... sandros has suggested we focus on the use cases 16:15:13 <AndyS> ... starting with Richard 16:15:31 <PatH> PatH has joined #rdf-wg 16:15:44 <PatH> Sorry Im late, having a problem phoning in. 16:16:22 <AndyS> ... email dec 21 at 13:43 UTC (??) 16:16:42 <AndyS> ... Richard, where is the key issues here? 16:17:20 <AndyS> sandro: for now, just think about UC, not analysis one UC until several looked at 16:17:35 <AndyS> ... and we may see the common elements 16:18:01 <AndyS> david: need to look sufficiently deeply 16:18:19 <Zakim> +PatH 16:18:22 <Zakim> -cygri 16:18:49 <AndyS> cygri: this UC was about the exchange of ... really 2 UC ... and a Higgs boson was seen 16:20:25 <AndyS> sandro: what are other people hoping to get out of NG 16:20:44 <Guus> q+ to suggest BBC one to take on, if it takes some time for Richard to come back 16:20:55 <AndyS> david: Jeremy suggested a text case approach 16:21:02 <PatH> maybe if i leave...? 16:21:05 <AndyS> ... will get to that on the hour 16:21:30 <PatH> pchampi, not knowingly. 16:21:51 <JeremyCarroll> JeremyCarroll has joined #rdf-wg 16:23:01 <AndyS> Jeremy: previously, have found that test cases stress the common ground, not the conceptualisms around it 16:23:10 <PatH> +1 to jeremy 16:23:16 <Guus> +1 16:23:17 <PatH> zakim, mute me 16:23:17 <Zakim> PatH should now be muted 16:23:42 <Zakim> +??P8 16:23:50 <cygri> zakim, ??P8 is me 16:23:50 <Zakim> +cygri; got it 16:23:58 <AndyS> ... I suggested on test case. We can formally resolve on that then move to next test case. 100 small steps. 16:24:10 <AndyS> (link please) 16:24:13 <PatH> richard is back :-) 16:24:38 <Guus> +1 to the ill-formedness of that test case 16:24:53 <gavinc> { ( {a:b c:d e:f}, mailto:ivan@w3.org ), ( {}, mailto:ivan@w3.org ) } 16:25:16 <Guus> q+ to syggest to do the strawpoll first 16:25:38 <JeremyCarroll> http://lists.w3.org/Archives/Public/public-rdf-wg/2011Dec/0123.html 16:25:41 <davidwood> ack Guus 16:25:41 <Zakim> Guus, you wanted to suggest BBC one to take on, if it takes some time for Richard to come back and to syggest to do the strawpoll first 16:26:03 <AndyS> q+ 16:26:04 <gavinc> STRAWPOLL: RDF 1.1 Recommendation will not recommend the use { ( {a:b c:d e:f}, mailto:ivan@w3.org ), ( {}, mailto:ivan@w3.org ) } 16:26:42 <AndyS> q+ to ask if this is syntax or abstract data model? 16:26:53 <davidwood> ack AndyS 16:26:53 <Zakim> AndyS, you wanted to ask if this is syntax or abstract data model? 16:27:28 <AndyS> AndyS: Is this syntax? 16:27:33 <AndyS> Jeremy: Yes. 16:27:38 <PatH> lets stick to abstract model first. 16:27:40 <Guus> it is not TriG syntax, btw 16:27:59 <pfps> q+ 16:28:06 <davidwood> q? 16:28:13 <davidwood> ack pfps 16:28:25 <AndyS> pfps: is this an RDF dataset? 16:28:46 <Zakim> +Eric 16:28:49 <AndyS> jeremy: syntax wrong -it is an RDF dataset that has a repeated label on two graphs 16:29:21 <cygri> RDF Dataset definition includes: "Graph names are unique within an RDF dataset." 16:29:31 <AndyS> pfps: RDF dataset has a label only once. 16:29:36 <cygri> so, that's a negative test case. 16:30:28 <gavinc> https://gist.github.com/1506664 16:30:31 <davidwood> q? 16:30:34 <gavinc> Not A Dataset https://gist.github.com/1506664 16:30:39 <PatH> Jeremy is trying to locate the shore by first putting a pole clearly in the water. 16:31:05 <AndyS> david: Negative test, concept, not synatx, 16:31:18 <JeremyCarroll> STRAWPOLL: RDF 1.1 Recommendation will not recommend the use { ( {a:b c:d e:f}, mailto:ivan@w3.org ), ( {}, mailto:ivan@w3.org ) } 16:31:33 <cygri> +1 16:31:35 <yvesr> +1 16:31:40 <MacTed> (will not recommend? or will disrecommend?) 16:31:41 <JeremyCarroll> +1 16:31:41 <Guus> +1 16:31:44 <davidwood> +1 16:31:46 <pchampin> +1 16:31:52 <gavinc> +1 16:31:53 <davidwood> Sandro: +1 16:31:53 <AZ> +1 16:31:53 <AndyS> +1 16:31:55 <Guus> we can formulate this stronger, I think 16:31:56 <ericP> +1 16:31:56 <pfps> +1 to what david is saying, not to the strawpoll per se 16:32:00 <MacTed> +1 for not recommending ... but I think that's not the right expression 16:32:07 <AlexHall> +1 16:33:03 <pfps> STRAWPOLL: In an RDF dataset, there must be a function (and not a general relation) from "tags" to graphs 16:33:09 <ericP> q? 16:33:16 <AndyS> guus: test suite 16:33:29 <ericP> i think a test suite would be bound to a serialization 16:33:33 <AndyS> ... we need a owner/manager 16:33:44 <AndyS> q+ 16:34:18 <AndyS> david: if we can't get a test case, keep proposal and not loose it 16:34:21 <gavinc> PROPOSAL : In a TriG document a graph IRI must not be used to label more then one graph. 16:34:51 <AndyS> q- 16:34:57 <Guus> proposal to use TriG as test-suite serialization language 16:35:00 <PatH> English version: the WG will not recommend any construction in which two different graphs are assigned the same label. 16:35:26 <PatH> zakim, unmute me. 16:35:26 <Zakim> PatH should no longer be muted 16:35:30 <AndyS> why not have "tests" being text files for now, pending syntax? 16:35:34 <ericP> a SPARQL-style test suite of a set of a mapping of turtle docs to unique graph names (and default graph) would capture this test without steaking out an interpretation of TriG 16:35:56 <davidwood> PROPOSAL : In a TriG document or similar syntax constituting an RDF dataset, a graph IRI must not be used to label more then one graph. 16:36:02 <AndyS> +1 to ericP as SPARQL tests can write these down 16:36:27 <cygri> q+ 16:36:28 <PatH> steaking out, nice. we can try roasting out as well. 16:36:34 <davidwood> ack cygri 16:36:46 <PatH> sounds much better 16:36:56 <PatH> zakim, mute me. 16:36:56 <Zakim> PatH should now be muted 16:37:02 <AndyS> cygri: option for different relationships of a "name" and a graph 16:37:17 <AndyS> ... may or may not be mixed in an RDF dataset 16:37:46 <AndyS> ... ??allow the name multiple times for different relationships 16:38:01 <JeremyCarroll> q+ to speak against Richard 16:38:10 <gavinc> <mailto:ivan@w3.org> <relation1> { 16:38:11 <gavinc> <a> <b> <c> . 16:38:13 <gavinc> } 16:38:14 <gavinc> <mailto:ivan@w3.org> <relation2> { 16:38:16 <gavinc> 16:38:17 <gavinc> } 16:38:23 <AndyS> <labeluri> :p1 {} . <labeluri> :p2 {} 16:38:25 <ericP> <G1> :logSemantics { <s1> <p1> <o1> } ; :signs { <s2> <p2> <o2> } 16:38:39 <davidwood> ack JeremyCarroll 16:38:39 <Zakim> JeremyCarroll, you wanted to speak against Richard 16:38:47 <MacTed> tweaked PROPOSAL : Within a single RDF dataset, whether expressed by a TriG document or any other syntax, a single graph IRI (1) MUST NOT be used to label more then one graph; (2) MAY ONLY be used to label one graph. 16:38:58 <Zakim> -cygri 16:39:03 <AndyS> Jeremy: we can revise if we decide that. Test case is more concrete probe. Not too many choices. 16:39:15 <Zakim> +??P8 16:39:18 <cygri> zakim, ??P8 is me 16:39:18 <Zakim> +cygri; got it 16:39:18 <PatH> i see what richard is saying, but agree with jeremy 16:39:55 <PatH> any decision will rule out some possibilities. 16:40:14 <MacTed> intentionally so 16:40:22 <PatH> are they different? 16:41:06 <Guus> PROPOSAL Within a single RDF dataset MUST NOT be used to label more then one graph 16:41:07 <davidwood> PROPOSAL : Within a single RDF dataset, regardless of syntax, a single graph IRI (1) MUST NOT be used to label more then one graph; (2) MAY ONLY be used to label one graph. 16:41:19 <PatH> q+ 16:41:20 <MacTed> (1) and (2) are intentionally redundant. the first conforms more to RFC grammar; the second is more clear to many readers. 16:41:27 <AndyS> Jeremy: multiple labelled blocks is still one graph is a possibility 16:41:27 <ericP> PROPOSAL : Within a single RDF dataset, regardless of syntax, a single graph IRI (1) MUST NOT be used to label more then one graph 16:41:31 <davidwood> ack PatH 16:41:34 <PatH> zakim, unmute me 16:41:34 <Zakim> PatH was not muted, PatH 16:42:19 <AndyS> PatH: one reason for not doing this is where identity is expensive 16:42:20 <cygri> we just need to detect identity of the URIs, not of the graphs 16:42:36 <PatH> zakim, mute me 16:42:36 <Zakim> PatH should now be muted 16:42:57 <davidwood> PROPOSAL : Within a single RDF dataset, regardless of syntax, a single graph IRI MUST NOT be used to label more then one graph 16:43:06 <ericP> +1 16:43:06 <MacTed> +1 16:43:07 <pfps> +1 16:43:10 <davidwood> Sandro: +1 16:43:11 <Guus> +1 16:43:11 <davidwood> +1 16:43:12 <PatH> +1 16:43:12 <gavinc> +1 16:43:15 <cygri> +1 16:43:15 <AZ> +1 16:43:15 <AlexHall> +1 16:43:18 <pchampin> +1 16:43:22 <yvesr> +1 16:43:22 <AndyS> +1 16:43:57 <AndyS> RESOLVED: Within a single RDF dataset, regardless of syntax, a single graph IRI MUST NOT be used to label more then one graph 16:44:00 <PatH> cygri, yes of course. sorry. 16:44:24 <AndyS> david: The use case ... richard 16:44:41 <AndyS> cygri: UC about web crawling and crawling RDF data 16:45:02 <AndyS> ... at the time of crawl may not know the use to be made of the data 16:45:12 <AndyS> ... i.e. crawler grabs stuff 16:45:24 <AndyS> ... so keep things in separate graphs 16:45:29 <AndyS> ... keep source info 16:46:22 <AndyS> ... at this stage, data considered is by source only, nothing outside 16:46:28 <Guus> remark about the resolution: you can interpret it as an implicit resolution that graph names must be IRIs. Should we make this explicit? 16:46:40 <AndyS> ... important - store this in a SPARQL store for query using GRAPH 16:47:15 <AndyS> ... important - crawls can be exchanged (practical reasons) - nquads where 4th slot is the source 16:47:22 <AndyS> ... source URL 16:47:49 <AndyS> david: Are you using the source URL as the graph URI? 16:47:59 <davidwood> q? 16:48:31 <AndyS> cygri: In several examples, inc billion triples dump, and Sindice, then source URL is 4th slot in the dump 16:48:43 <AndyS> ... makes it easy to understand 16:49:17 <AndyS> sandro: similar : tabulator and ?? but they use tag as a retriveal event 16:49:35 <AndyS> ... if higher costs are acceptable 16:49:47 <PatH> do these systems use RDF metadata in which the graph identifiers are used to identify the graphs? 16:50:20 <cygri> q+ to answer pat 16:50:30 <AndyS> david: how does this inform you? 16:51:25 <AndyS> cygri: to PatH: they refer to the source, not the graph 16:51:26 <PatH> great. 16:51:59 <AndyS> cygri: does not make much sense about the merge of all the graphs 16:52:34 <AndyS> ... or consistency. It data management to keep them separate. May subset and do other stuff later. 16:52:35 <PatH> that really is very nice, and allows an elegant semantic 'gloss' to account for this being conformant. 16:53:08 <AndyS> ... semantics is "not good" lots of things wrong and need to be sorted out by the app on top later 16:53:23 <AndyS> ... we want a way to split the triples by source 16:53:26 <PatH> q+ 16:53:35 <PatH> zakim, unmute me 16:53:35 <Zakim> PatH should no longer be muted 16:53:54 <davidwood> ack cygri 16:53:54 <Zakim> cygri, you wanted to answer pat 16:54:06 <davidwood> ack PatH 16:54:40 <Zakim> -cygri 16:54:40 <AndyS> PatH: The sticking point for me has been the way URIs are used in difefrent ways - now clear in Richards UC that this is not the case 16:54:55 <Zakim> +??P8 16:55:00 <cygri> zakim, ??P8 is me 16:55:00 <Zakim> +cygri; got it 16:55:01 <ericP> AggregateDB GETs <http://a.example/doc1.ttl> at with headers <H>, could create any of these DBs: 16:55:04 <ericP> snapshot: <http://a.example/doc1.ttl>: { <s1> <p1> <o1> } 16:55:07 <ericP> with-metadata: default graph: { <http://a.example/doc1.ttl> http:headers <H> } 16:55:09 <AndyS> ... so we can add metadata etc 16:55:09 <ericP> <http://a.example/doc1.ttl>: { <s1> <p1> <o1> } 16:55:12 <ericP> with history: default graph: { </2011-12-21T11:50/http://a.example/doc1.ttl> http:resource <http://a.example/doc1.ttl> ; http:headers <H> } 16:55:15 <ericP> </2011-12-21T11:50/http://a.example/doc1.ttl>: { <s1> <p1> <o1> } 16:55:19 <ericP> does that capture likely designs? 16:55:28 <AndyS> s/semantics/data/ 16:55:49 <AndyS> PatH: Will write up my point. 16:56:27 <PatH> did we lose richard again? 16:56:34 <ericP> q+ 16:56:42 <davidwood> zakim, who is talking? 16:56:42 <AndyS> scribe+ 16:56:54 <Zakim> davidwood, listening for 10 seconds I heard sound from the following: Sandro (64%), Eric (14%) 16:57:11 <PatH> zakim, mute me 16:57:11 <Zakim> PatH should now be muted 16:57:23 <MacTed> scribe was breaking up 16:57:30 <Zakim> -AndyS 16:57:42 <AndyS> Oh dear 16:58:33 <PatH> scrumpy rules. 16:58:36 <Zakim> +[IPcaller] 16:58:46 <AndyS> zakim, IPcaller is me 16:58:46 <Zakim> +AndyS; got it 16:58:48 <PatH> noise... 16:58:54 <JeremyCarroll> Zakim, mute AndyS 16:58:54 <Zakim> AndyS should now be muted 16:59:30 <AndyS> ericP: try to cover 3 cases 16:59:40 <AndyS> ... stores in RDF DB 16:59:49 <AndyS> ... store in doc 1 and add header to dft graph 16:59:55 <davidwood> EricP: See his syntax above, where someone GETs graphs. 17:00:11 <cygri> it covers mine 17:00:12 <AndyS> ... store in doc1 in an dft doc nfo on this 17:00:25 <AndyS> I am scribing 17:00:40 <PatH> it covers enough to give rise to much discussion... 17:01:00 <AndyS> sandro (is unclear to the scribe) 17:01:15 <Zakim> -Sandro 17:01:36 <PatH> sandro: may be other cases. 17:01:42 <davidwood> EricP has covered the two variants of the use case as I have seen them. 17:01:55 <Zakim> +Sandro 17:01:58 <AndyS> ack me 17:02:01 <ericP> ack me 17:02:22 <AndyS> I'll pass 17:02:36 <davidwood> ack Sandro 17:02:38 <Zakim> -AndyS 17:02:50 <ericP> scribenick: ericP 17:03:22 <ericP> sandro: [silence] 17:03:58 <Zakim> +[IPcaller] 17:04:01 <davidwood> Zakim must be forked. I hear some. 17:04:03 <AndyS> zakim, IPCAller is me 17:04:03 <Zakim> +AndyS; got it 17:04:08 <PatH> Zakim is getting tired. 17:04:23 <AndyS> david: bad conditions 17:04:38 <AndyS> ... regards for the holidays 17:04:43 <AndyS> ADJOURNED 17:04:47 <yvesr> bye! 17:04:47 <pchampin> thanks 17:04:48 <Zakim> -davidwood 17:04:49 <AndyS> next meeting Jan 4. 17:04:50 <Zakim> -Sandro 17:04:50 <Zakim> -JeremyCarroll 17:04:51 <Zakim> -MacTed 17:04:52 <Guus> bye 17:04:53 <Zakim> -yvesr 17:04:54 <AZ> bye 17:04:54 <PatH> eric, can you email that example? 17:04:55 <Zakim> -Peter_Patel-Schneider 17:04:56 <cygri> thanks david! 17:04:59 <Zakim> -AndyS 17:04:59 <Zakim> -gavinc 17:05:01 <Zakim> -pchampin 17:05:02 <Zakim> -cygri 17:05:05 <Zakim> -Eric 17:05:10 <Zakim> -AlexHall 17:05:14 <Zakim> -AZ 17:05:15 <Zakim> -Scott_Bauer 17:05:16 <PatH> muted merry Xmas to all. 17:05:16 <AlexHall> AlexHall has left #rdf-wg 17:06:02 <Zakim> -Guus # SPECIAL MARKER FOR CHATSYNC. DO NOT EDIT THIS LINE OR BELOW. SRCLINESUSED=00000391