14:28:13 RRSAgent has joined #dawg 14:28:13 logging to http://www.w3.org/2005/09/20-dawg-irc 14:28:19 zakim, this will be dawg 14:28:19 ok, ericP, I see SW_DAWG()10:30AM already started 14:28:28 RRSAgent, pointer? 14:28:28 See http://www.w3.org/2005/09/20-dawg-irc#T14-28-28 14:28:33 zakim, who is here? 14:28:33 On the phone I see PatH 14:28:34 On IRC I see RRSAgent, Zakim, patH, franconi, LeeF, AndyS, jeen, SerT, SteveH, afs, DanC, ericP 14:28:42 Meeting: DAWG 14:29:27 +??P18 14:29:30 Chair: PatH 14:29:37 Scribe: SteveH 14:29:38 +[IBMCambridge] 14:29:49 Time: 14:30Z 14:29:50 Zakim, [IBMCambridge] is LeeF 14:29:50 +LeeF; got it 14:30:05 Zakim, ??P18 is JeenB 14:30:05 +JeenB; got it 14:30:08 +Ericp 14:30:11 Yoshio has joined #dawg 14:30:52 +??P9 14:30:55 zakim, ??P9 is AndyS 14:30:55 +AndyS; got it 14:31:14 zakim, who is on the phone? 14:31:14 On the phone I see PatH, JeenB, LeeF, Ericp, AndyS 14:31:16 +Yoshio 14:31:17 kendall has joined #dawg 14:31:36 +??P8 14:32:03 +Kendall_Clark 14:32:35 zakim, ??P8 is Souri 14:32:35 +Souri; got it 14:34:06 +Ernesto_Damiani 14:34:07 Btw, I'm empowered to say (:>) that Bijan will be here shortly; he had a dr's appointment thing to take care of. 14:34:19 All : Please join the phone telecon. 14:34:35 That is where we will be discussing this material. 14:34:51 +NickG 14:35:02 Zakim, NickG is SteveH 14:35:02 +SteveH; got it 14:35:24 JosD has joined #dawg 14:35:53 zakim, Ernesto_Damiani is really Enrico_Franconi 14:35:53 +Enrico_Franconi; got it 14:36:15 +Jos_De_Roo 14:36:48 Let's hope the real (if there are such) "Voices from Above" are more helpful, coherent, and friendly. :> 14:39:34 excuse me 14:39:52 franconi: the intent is to give a formal semantics for SPARQL as-is 14:40:04 Erico: posted a semantics that are intened to be consistent wiht the spirit of SPARQL and resolving the subgraph vs. entailment debate 14:40:18 http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JulSep/0450.html 14:40:26 s/Erico/Enrico/ 14:40:28 + +44.797.160.aaaa 14:41:11 zakim, aaaa is Ian_Horrocks 14:41:11 +Ian_Horrocks; got it 14:41:17 SP? 14:41:47 I'm hearing a lot of noise on the call. 14:41:49 hmm, can we mute Ian? 14:42:03 +[UMD] 14:42:06 zakim, please mute Ian_Horrocks 14:42:06 Ian_Horrocks should now be muted 14:42:14 Bijan, i presume? 14:42:21 yes, I assume so 14:42:35 bijan has joined #dawg 14:42:49 zakim, UMD is Bijan 14:42:49 +Bijan; got it 14:43:16 zakim, please unmute Ian_Horrocks 14:43:16 Ian_Horrocks should no longer be muted 14:43:18 hmm, yes, someone should at least tell Ian that he's muted 14:43:41 *6 i believe 14:43:58 Maybenot :) 14:44:15 zakim, please mute Ian_Horrocks 14:44:15 Ian_Horrocks should now be muted 14:45:45 SteveH, if you hear anything you'd like in the record, you're listed as scribe 14:46:02 ericP, yeah, I cant think and type though :-/ 14:46:12 I'm trying to scribe, since I can't think anyway. :> 14:46:17 also, things you particularly don't want in the record, they're probably relevent too 14:46:27 moreso i suspect 14:47:14 Enrico: 14:47:16 match only URIs and literals 14:47:33 franconi, skolemise bNodes by replacing with newly minted bNodes gets you the semantics we have now (?) 14:47:34 skolemize bNodes into URIs 14:47:44 Skolomise bNodes, do entailment (no bNodes), deskolomize 14:47:54 and convert them to bNodes [upon reporting?] 14:48:00 . 14:48:21 Is this the same as saying 1/ no shared bNodes and 2/ only bnodes from the original graph in solutions? 14:49:58 zakim, q+ 14:49:58 I see bijan on the speaker queue 14:50:37 franconi: SPARQL can be represented by an algebra (like SQL) 14:52:02 bijan: I'd like to have a calculus for all of SPARQL. But we don't have that, and that's not what Enrico has proposed. 14:53:09 q? 14:53:12 ack bijan 14:53:58 zakim, please unmute Ian_Horrocks 14:53:58 Ian_Horrocks should no longer be muted 14:54:27 zakim, please mute Ian_Horrocks 14:54:27 Ian_Horrocks should now be muted 14:57:49 franconi: our proposal doesn't include CONSTRUCT semantics 14:58:04 Enrico: CONSTRUCT is analogous to object construction in SQL 14:58:26 Enrico: this is not covered by my proposal 15:02:24 franconi: two equivalent graphs may give different answers (lean v's fatty) 15:02:46 q+ 15:03:30 franconi: if we add a minimisation step to skolemise, unskolemise then we get same results 15:04:51 franconi: minimisation is expensive, so should be optional 15:05:08 (NP hard (?) in worst case) 15:06:04 I'm unclear as to NP-hard WRT what? The queried data or the result set only. 15:06:16 franconi: minimisation is only neccesary where there are bNodes in the data 15:06:54 i guess graph match is NP-hard. seems it would be that NP prob, but i can't quite see where it applies 15:07:33 AndyS: If I understand Enrico correctly, the minimisation can occur outside of the graph matching, on the result set only, which would mean it's NP-Hard in the size of the result set. But, I give no guarantees that I understand any of this correctly. :-/ 15:07:35 u can see minimisation as conjunctive query containment 15:07:50 zakim, q+ 15:07:50 I see AndyS, bijan on the speaker queue 15:08:51 zakim, q- 15:08:51 I see AndyS on the speaker queue 15:09:03 LeeF: I hope that is true but somethign said didn't back that up. It seems to be equiv to entailment WRT the data. 15:09:38 zakim, q+ to say that there are two issues: 1) the specification of the correct answers to a query given a semantics and 2) how, in the language or protocol, which semantics are desired or in force 15:09:38 I see AndyS, bijan on the speaker queue 15:09:53 q+ 15:10:37 q+ to say that it might be a feature of the service - not a client option. 15:11:02 agh! 15:11:21 ack AndyS 15:11:21 AndyS, you wanted to say that it might be a feature of the service - not a client option. 15:11:42 result set 15:11:45 AndyS: the mini. alg. NPH w.r.t what? result set or KB size 15:12:02 q+ Ian_Horrocks a reminder to poll him 15:12:03 franconi: w.r.t. coreferent nodes 15:12:10 franconi: NP-hard WRT to coreferenced nodes 15:12:12 q+ Ian_Horrocks to say a reminder to poll him 15:12:52 coreferent variables 15:14:04 -Ian_Horrocks 15:14:19 q-Ian_Horrocks 15:14:48 Ian, got that. 15:15:34 ack bijan 15:15:34 bijan, you wanted to say that there are two issues: 1) the specification of the correct answers to a query given a semantics and 2) how, in the language or protocol, which 15:15:37 ... semantics are desired or in force 15:15:40 ack JoseD 15:15:46 ack JosD 15:16:23 JosD: suppose we have _:C . _:A _:B _:C 15:16:40 JosD: _:c minimizes to _:a _:b _:c ? 15:16:56 zakim, mute me 15:16:56 Kendall_Clark should now be muted 15:17:07 Enrico: this minimizes to _:c 15:17:12 _:C . _:A 15:17:55 Enrico: this [above] is already minimized 15:18:49 minimisation interacts with FILTER 15:19:05 explain? 15:19:20 q+ 15:19:21 FILTER(bNode(?a))? 15:19:41 u can filter out tuples on the basis of blank nodes 15:20:07 so FILTER can throw away ground tuples 15:20:20 yep 15:20:22 q+ 15:20:27 tx 15:20:38 ack bijan 15:20:56 ack AndyS 15:21:51 PatH: i think we want a flag in the query to say "i want a minimal solutin" 15:22:04 Enrico: this is "distinct" in SQL 15:22:09 franconi: its equivalent to SQLs DISTINCT 15:22:25 I dont see how its related to SQLs DISTINCT 15:22:28 ... UNION may add redundancy 15:22:42 It's more expensive than DISTINCT because DISTINCT is streamable 15:22:56 ... same with sub select [did i get that right?] 15:23:38 Yes - I think it is the same 15:23:41 AndyS...that's an interesting point 15:23:49 q+ 15:24:03 hmm, in my ample personal experience (alas), Enrico is nothing like a fundamentalist Christian. :> 15:26:56 zakm, ack bijan 15:27:01 zakim, ack bijan 15:27:01 I see no one on the speaker queue 15:27:31 q+ to ask why if this entailment parameter could be put in the service or the graph 15:30:43 q+ to say that not handling OWL explicitly is not the same as blocking possible OWL-enabling 15:32:28 q+ 15:32:37 q- 15:32:47 http://www.w3.org/2003/12/swa/dawg-charter#derivedGraphs [[ 15:32:49 The working group must recognize that RDF graphs are often constructed by aggregation from multiple sources and through logical inference, and that sometimes the graphs are never materialized. Such graphs may be arbitrarily large or infinite. 15:32:53 ]] 15:33:57 I'd point out "never materialized" in the talk of infinite graphs and inf closures 15:34:01 zkim, ack EricP 15:34:04 zakim, ack EricP 15:34:04 ericP, you wanted to ask why if this entailment parameter could be put in the service or the graph 15:34:07 I see franconi on the speaker queue 15:34:08 no prob, you said it better than I could :) 15:34:47 bijan, and thanks for waving the RDFS flag :) 15:34:57 no problem 15:35:56 eric: is making the entailement level a feature of the service acceptable to Bijan and Enrico? 15:36:03 Bjian, Enrico: Yes 15:36:17 Bijan: must stil define the outcome of various entailment levels 15:42:41 q- 15:43:05 IanH has joined #dawg 15:44:12 Sorry I dropped off the call - ran out of battery. 15:44:26 we forgive you. 15:45:43 zakim, unmute me 15:45:43 Kendall_Clark should no longer be muted 15:46:02 Bijan: how to move forward? 15:47:10 Test cases needed 15:47:18 Use cases and test cases would useful. 15:48:18 Everyone agrees that SPARQL should not be req'd to respect RDF entailment. 15:48:26 (is that correct scribing?) 15:48:39 kendall: yes 15:48:49 seems worth recording right :> 15:48:55 Yes. 15:49:09 Everyone also agrees that SPARQL should be *able* to respect RDF entailmetn :) 15:49:11 ? 15:49:12 and sometimes celebrating 15:49:59 "should" yes - "must" (i.e. every impl must provide that mode) then, no 15:50:22 I.e., SPARQL should specify what the correct result sets are wrt to both "no" entailment and rdf entailment (and presuambly simple entailment) 15:50:24 patH (?) consider the query: Do there exist any rdfs:ContainerMembershipProperty's against an empty graph - services should not be required to give any answers to that, even though logically they should 15:50:46 Bijan - agree 15:51:39 bijan, "specify"? wouldn't simply referring to relevant parts of RDF Semantics be enough? 15:51:51 jeen, no, because of things like minimality 15:52:41 Enrico: URL of service extended to include the minimalization mode 15:52:49 I.e., if you say "subgraph of the rdfization of the dedeuctive closure", different implementation can give different answers depending on how it regards redundancy in the answers 15:52:54 Ian? Anythign to say? 15:52:54 zakim, who's on the phone? 15:52:54 On the phone I see PatH, JeenB, LeeF, Ericp, AndyS, Yoshio, Souri, Kendall_Clark, Enrico_Franconi, SteveH, Jos_De_Roo, Bijan 15:53:37 I missed a lot of the discussion, but what Enrico and Bijan are proposing sounds reasonable. 15:53:39 Who's gonna break the bad news to Connolly?! :> 15:53:48 "bad"? 15:54:00 precisely 15:54:12 We have a way forward - not a blockage 15:55:22 boo hiss 15:55:26 static on the line 15:55:46 -JeenB 15:56:29 jeen, that's ok, we'd just gotten to the hippi love fest 15:56:41 oh boo, I missed the group hug :( 15:57:27 q+ to ask if we have agreed one way or the other on whether this will affect the protocol (clients specifyoing desired entailment) 15:57:59 well, i want to know if we *have* to. that didn't seem clear. 15:58:22 Right, that as well. 15:58:36 :> 15:58:45 AndyS, what would be helpful is a list of the things that change and what doesn't. we've had a good discussion today. we need to log that discussion 16:00:05 ACTION Bijon: take a pass through the editor's draft, listing what will change with the new semantics understanding, and what will not 16:00:11 ACTION Enrico: take a pass through the editor's draft, listing what will change with the new semantics understanding, and what will not 16:00:14 zakim, mute me 16:00:14 Kendall_Clark should now be muted 16:00:32 s/Bijon/Bijan 16:00:33 Who is this Bijon dude? 16:00:49 ACTION Bijan: take a pass through the editor's draft, listing what will change with the new semantics understanding, and what will not 16:00:53 action -1 16:01:08 Who's this Bijon person? 16:01:23 your good twin? 16:01:26 Action Pat to summarize the consensus. 16:01:36 I've been asked many times to leave conference talks because I was typing too loud! 16:01:42 One procedural thing - we are having a meeting next week (27/Sept) aren't we? Hint, hint 16:01:44 I feel (or hear?) Pat's pain 16:01:55 ACTION Pat: summarize the entailment consensus 16:02:20 I'll scribe 16:02:31 Thx Kendall for offerig to scribe 16:02:42 I have to go 2 minutes ago. Ciao! 16:02:48 -Souri 16:02:50 -Jos_De_Roo 16:02:52 -SteveH 16:02:54 -Kendall_Clark 16:03:02 next meeting: next week, chair: DanC or ericP, scribe: KendallC 16:03:02 -Yoshio 16:03:09 q- 16:03:37 ADJOURN 16:03:53 Andy asks about whether skolomisation of bNodes and equiv to restricting to told facts. 16:03:59 rrsagent, please draft minutes 16:03:59 I have made the request to generate http://www.w3.org/2005/09/20-dawg-minutes.html ericP 16:04:09 rrsagent, please make minutes world-readable 16:04:09 I'm logging. I don't understand 'please make minutes world-readable', ericP. Try /msg RRSAgent help 16:04:12 rrsagent, please make minutes world readable 16:04:12 I'm logging. I don't understand 'please make minutes world readable', ericP. Try /msg RRSAgent help 16:04:17 It is confirmed. 16:05:28 zakim, please leave us 16:05:28 I don't understand 'please leave us', ericP 16:05:34 zakim, bye 16:05:34 leaving. As of this point the attendees were PatH, LeeF, JeenB, Ericp, AndyS, Yoshio, Kendall_Clark, Souri, SteveH, Enrico_Franconi, Jos_De_Roo, +44.797.160.aaaa, Ian_Horrocks, 16:05:34 Zakim has left #dawg 16:05:37 ... Bijan 16:05:44 rrsagent, bye 16:05:44 I see 3 open action items saved in http://www.w3.org/2005/09/20-dawg-actions.rdf : 16:05:44 ACTION: Enrico to take a pass through the editor's draft, listing what will change with the new semantics understanding, and what will not [2] 16:05:44 recorded in http://www.w3.org/2005/09/20-dawg-irc#T16-00-11 16:05:44 ACTION: Bijan to take a pass through the editor's draft, listing what will change with the new semantics understanding, and what will not [3] 16:05:44 recorded in http://www.w3.org/2005/09/20-dawg-irc#T16-00-49 16:05:44 ACTION: Pat to summarize the entailment consensus [4] 16:05:44 recorded in http://www.w3.org/2005/09/20-dawg-irc#T16-01-55