From SPARQL Working Group
Revision as of 14:32, 15 April 2009 by Lfeigenb
Please justify/explain all edits to this page, in your "edit summary" text.
<LeeF> Present: Lee, Steve, Luke, Andy, ivanh, Orri, dnewman2, kasei, chimezie, john-l, ericp, Simon 13:51:19 <trackbot> Meeting: SPARQL Working Group Teleconference 13:51:21 <trackbot> Date: 14 April 2009 13:53:40 <LeeF> Regrets: Axel, Alex, Kjetil, Jacek 13:53:43 <LeeF> Chair: LeeF 13:54:47 <AndyS> Agenda AOB request - F2F1 - We seem to have 5 Boston, 7 Bristol. 13:56:31 <LeeF> AndyS, ack 13:56:38 <Zakim> SW_(SPARQL)10:00AM has now started 14:02:29 <ivanh> scribenick: ivanh 14:02:38 <ivanh> scribe: ivanh 14:03:24 <chimezie> zakim, unmute me 14:03:59 <ivanh> Topic: administravia 14:04:05 <kasei> zakim, mute me 14:04:05 <Zakim> kasei was already muted, kasei 14:04:09 <LeeF> -> http://www.w3.org/2009/sparql/wiki/Agenda-2009-04-14 agenda for today 14:04:24 <LeeF> PROPOSED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2009-04-07 14:04:26 <Zakim> -Lee_Feigenbaum 14:04:39 <Zakim> +EricP 14:04:51 <Zakim> +Lee_Feigenbaum 14:05:20 <Zakim> + +1.479.864.aaee 14:06:06 <ivanh> RESOLVED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2009-04-07 14:06:15 <ivanh> topic: next meeting 14:06:20 <ivanh> next tuesday, this time 14:06:32 <ivanh> regrets next week: orri, ivanh 14:06:55 <ivanh> lee: any report of our liaisons? 14:07:14 <ivanh> AndyS: i believe sparql and owl will ask us to review rdf:text 14:07:28 <AndyS> s/sparql/rif/ 14:07:38 <ivanh> LeeF: what we said that when the text is ready we will send it to the mailing list and see if anybody is interested 14:07:47 <ivanh> Topic: rechartering issue 14:08:47 <ivanh> lee: my understanding is that it should not affect our work at all 14:09:14 <ivanh> ... this may be naive or optimistic, but we should be able to do the same work the same way 14:10:24 <LeeF> ivanh: the way the charter is done today is extremely open-ended, and companies that really want to consider all their patent issues cannot make any patent disclosures right now since nothing specifies what will be part of a Recommendation 14:10:40 <LeeF> ... we received remarks from members making us consider this 14:11:13 <LeeF> ... we have sent out a rechartering proposal to the AC that charters this group to produce a non-Rec track document (WG Note) called features & rationale that lists all the features we want in SPARQL 14:11:18 <LeeF> ... formal lifespan is through the end of July 14:11:51 <LeeF> ... in practice, group keeps doing what it is doing, but needs to publish this WG Note before end of July 14:12:03 <chimezie> zakim, mute me 14:12:03 <Zakim> chimezie should now be muted 14:12:21 <LeeF> ... as soon as we publish a first public working draft of requirements (hopefully 2nd half of May), we would immediately do a 2nd rechartering that lists the features as part of the charter 14:12:31 <SteveH> q+ to ask about process and schedule 14:13:10 <LeeF> ... if new charter proposal is done by end of May, it would be official by end of June, so from beginning-to-mid of July the group could be active under the second charter 14:13:43 <LeeF> ... no problem if group begins to really work on specification work in May - the only thing the group should not do is publish a First Public Working Draft because that is what leads to patent issues 14:13:56 <LeeF> ... no problem with editors' drafts 14:14:24 <LeeF> ack SteveH 14:14:24 <Zakim> SteveH, you wanted to ask about process and schedule 14:15:17 <LeeF> SteveH: my understanding around Lee's intentions was to iterate on the process - consider other things as time permits 14:15:27 <LeeF> ... but now we no longer have that flexibility 14:15:47 <ericP> i don't think we'll be any more bound later on than we would be in any other WG 14:16:11 <LeeF> ... if charter is bound, we can't go ahead and talk about details because that would be outside the charter 14:17:07 <ericP> q+ to say that this ends up being like any other WG 14:17:48 <LeeF> LeeF: intention is to reach consensus on 'required' and 'time permitting' deliverables, and then include them all in the new charter 14:18:03 <AndyS> q+ to aks about IP review at WD stages 14:18:39 <kasei> is everyone else hearing lots of (mobile?) interference on the call, or is it just me? 14:18:52 <SimonS> me too. 14:19:41 <ivanh> ack ericP 14:19:41 <Zakim> ericP, you wanted to say that this ends up being like any other WG 14:20:06 <LeeF> ericp: we have patent policy since companies want to know what they're working on - the charter for the 2nd phase of the group ends up being like any other group 14:21:16 <ivanh> ack AndyS 14:21:16 <Zakim> AndyS, you wanted to aks about IP review at WD stages 14:21:47 <LeeF> AndyS: HP was quite well aware of this issue and noted that there is a call for IP disclosures when a FPWD is published 14:22:28 <LeeF> ivanh: Acknowledged; other groups are less flexible with respect to knowing potential IP issues 14:22:35 <LeeF> AndyS: will we all need to rejoin the WG? 14:22:39 <SteveH> wondering what the risk of not being rechartered is 14:23:04 <LeeF> ericP: I believe so. 14:23:40 <LeeF> AndyS: I don't think I could justify the new charter within my organization 14:31:05 <AndyS> s/I don't think I could justify the new charter within my organization/I don't think I could justify the nwe charter without internal review within my organsiation/ 14:24:29 <LeeF> ericP: I think for phase I we could continue the existing membership, but for phase II people would need to re-join 14:24:46 <SteveH> q+ 14:24:50 <LeeF> AndyS: if there's no grace period it will be very rough 14:25:23 <LeeF> AndyS: concern about publication - editors' drafts have been publicly available in the past - the difference between editors' draft being publicly available and FPWD is not that big 14:25:43 <LeeF> ivanh: difference is in commitment from WG and IP commitment 14:26:00 <LeeF> AndyS: I'm not sure if putting information in public exposes the editor's organization 14:27:05 <chimezie> zakim, mute me 14:27:05 <Zakim> chimezie was already muted, chimezie 14:27:07 <ivanh> ack SteveH 14:27:17 <LeeF> ivanh: if work is done on wiki as joint WG work, might not be an issue 14:27:43 <LeeF> SteveH: is there middle ground with a formal pause between 2 phases rather than a full rechartering? is this new standard operating procedure? 14:28:03 <Zakim> + +1.415.371.aaff 14:28:09 <LeeF> ivanh: this occurred because the initial charter was too broad 14:28:19 <LeeF> ... might have been better with initial work as a short-lived XG 14:29:14 <LeeF> SteveH: it's not a guarantee that phase II will happen (get rechartered 14:29:20 <LeeF> s/rechartered/rechartered)/ 14:29:28 <dnewman2> dnewman2 has joined #sparql 14:29:36 <LeeF> ivanh: yes, that's a risk, we tried to avoid it but it didn't work out 14:29:48 <LeeF> zakim, aaff is dnewman2 14:29:49 <Zakim> +dnewman2; got it 14:30:15 <LeeF> SteveH: what happens if the AC rejects the rechartering? 14:30:42 <LeeF> ericP: The Director has some leeway. Given the current state of affairs, I don't see this as a significant risk 14:36:01 <ericP> AndyS, SteveH, would a note to public-rdf-dawg would have make this better? 14:36:06 <SteveH> yes 14:36:30 <AndyS> No 14:37:07 <AndyS> Notifying current members before it went wide would have been helpful. Notify - not discuss. 14:37:23 <SteveH> yes, dicsussion is not neccesary 14:37:24 <ericP> (there's also the question of what volume of mutterings consitute time to alarm folks) 14:37:34 <ericP> s/consitute/consitutes/ 14:38:14 <ivanh> topic: features discussions 14:38:20 <ivanh> subtopic: parameterized inference <LeeF> summary: initial straw poll gives (+/0/-): 4/3/5 14:38:22 <LeeF> -> http://www.w3.org/2009/sparql/wiki/Feature:ParameterizedInference 14:38:26 <chimezie> zakim, unmute me 14:38:26 <Zakim> chimezie should no longer be muted 14:38:39 <ivanh> LeeF: there has been a bunch of discussions on that 14:38:54 <ivanh> chimezie, can you summarize? 14:39:25 <ivanh> chimezie: the link has a descriptioin for the general requirements 14:39:50 <ivanh> .. .last week we proposed the parametrization was used to describe the inference regime 14:40:00 <ivanh> ... this would go beyond owl 14:40:11 <ivanh> ... the wiki also mentions merging datasets 14:40:16 <LeeF> q+ to ask if merging datasets is the same as composite dataset issue 14:40:39 <ivanh> ... from ontologies or rules into the graph 14:40:51 <ivanh> ... that overlaps with the previous feature 14:41:06 <ivanh> ... there was a possibility to break this thing into a separate feature 14:41:28 <ivanh> ... the last thing is to parametrize whether an ent. regime would give all possible answers or not, for example 14:41:39 <ivanh> ... the general idea is to parametrize the possible answers 14:42:06 <ivanh> LeeF: is the topic of merging the same as composite datatypes 14:42:08 <AndyS> q+ to ask about protocol implications 14:42:12 <LeeF> s/datatypes/datasets 14:42:28 <ivanh> chimezie: yes, but if you want an additional answers, do you have to bring the data in, for example 14:42:31 <ivanh> q+ 14:43:07 <ivanh> LeeF: is it fair to say that the issue is giving all possible answers vs. not is a detail that can be worked out later, or has to discussed upfront 14:43:37 <SteveH> Zakim, who's talking 14:43:37 <Zakim> I don't understand 'who's talking', SteveH 14:43:38 <ivanh> chimezie: i do not think we have to answer this question now 14:43:50 <LeeF> ack leef 14:43:50 <Zakim> LeeF, you wanted to ask if merging datasets is the same as composite dataset issue 14:43:52 <LeeF> ack AndyS 14:43:53 <john-l> Zakim, please mute me 14:43:53 <Zakim> AndyS, you wanted to ask about protocol implications 14:43:55 <SimonS> q+ 14:43:56 <Zakim> john-l should now be muted 14:44:06 <ivanh> AndyS: this looks like describing the environment where the query executes 14:44:15 <ivanh> ... do we want that to be in the protocol, too 14:45:07 <ivanh> chimezie: i believe that the environment is for a set of query? 14:45:35 <ivanh> AndyS: al lthe examples have a syntax for replacing datasets 14:45:36 <Zakim> On IRC I see dnewman2, ericP, LukeWM_, SteveH, ivanh, SimonS, Zakim, RRSAgent, chimezie, kasei, AndyS, AndyS_, LeeF, trackbot, john-l 14:45:41 <LeeF> zakim, mute aaee 14:45:41 <Zakim> +1.479.864.aaee should now be muted 14:45:48 <ivanh> ... could one consider protocol entries instead of the query 14:45:55 <SimonS> q- 14:46:01 <ivanh> chimezie: you mean being in the http request body? 14:46:26 <ivanh> AndyS: we have default graph uri in the query, something like that ought to be considered 14:46:28 <LeeF> res discussions 14:46:47 <LeeF> q? 14:46:47 <ivanh> chimezie: it would be a similar request adn we should probably consider that 14:46:59 <LeeF> ack ivanh 14:47:27 <AndyS> I see two features - inference control and dataset composition. Related but different. 14:47:45 <LeeF> AndyS, I agree with you - there is also service description, but we've considered that on its own already 14:47:47 <SimonS> ack. 14:48:22 <AndyS> Yes - service description is the service controlling things - this is query controlling things. 14:48:34 <LeeF> AndyS, agreed. 14:49:34 <ivanh> LeeF: i see two different things for the 'vote', inference control and dataset composition 14:49:55 <chimezie> zakim, mute me 14:49:55 <Zakim> chimezie should now be muted 14:50:07 <ivanh> let us do a sraw poll on the inference control/description 14:50:45 <SimonS> +q 14:50:49 <ivanh> ... whether it is on the protocol or the language level are details, we should have a poll on whether this particular feature is to be considered 14:50:54 <LeeF> ack SimonS 14:51:08 <chimezie> zakim, unmute me 14:51:08 <Zakim> chimezie should no longer be muted 14:51:12 <ivanh> SimonS: i think this is useful, but it is very fuzzy, there are so many different things one could do 14:51:23 <ivanh> ... we need clarification before working on the details 14:51:47 <ivanh> ... i have the feeling that we are mixing different things here, bringing in rules for example 14:51:56 <ivanh> ... in principle I think this is useful 14:52:12 <ivanh> LeeF: any other comments? 14:52:14 <ivanh> .... 14:52:16 <ivanh> .... 14:52:20 <SteveH> -1, too early, not enough experience 14:52:21 <ericP> -1 # fear it would distract from my highest priorities: update and lists 14:52:33 <kasei> 0 14:52:35 <ivanh> +1 14:52:35 <john-l> 0 14:52:36 <chimezie> +1 extends usefulness of SPARQL 14:52:40 <dnewman2> +1 useful feature 14:52:51 <AndyS> -1, too early not enough implementation experience. Use extensions 14:53:01 <chimezie> zakim, mute me 14:53:02 <Zakim> chimezie should now be muted 14:53:06 <LeeF> Orri: +1 (will not be trivial, but important) 14:53:09 <SimonS> -1 useful, but too fuzzy. Other priorities 14:53:12 <LeeF> 0 14:53:23 <Zakim> - +1.479.864.aaee 14:53:31 <LukeWM_> -1 seems vague at the moment 14:53:43 <ivanh> q+ 14:54:03 <chimezie> zakim unmute me 14:54:04 <Zakim> + +1.479.864.aagg 14:54:36 <SteveH> q+ 14:54:43 <ivanh> ack ivanh 14:54:45 <AndyS> q+ 14:55:01 <chimezie> q+ to respond to LeeF 14:55:28 <dnewman2> +q 14:55:32 <LeeF> ack SteveH 14:55:44 <ivanh> SteveH: people have been doing this for some time without this feature 14:55:45 <LeeF> ivanh: how is SPARQL/OWL useful without this? 14:56:08 <ivanh> ... people have been doing that for a while 14:56:16 <AndyS> ack AndyS 14:56:26 <LeeF> LeeF: there are other ways (rather than specifying in the query) to know that you are querying an endpoint that does SPARQL/OWL 14:56:31 <ivanh> AndyS: much the same thing as SteveH.. 14:58:31 <LeeF> ack chimezie 14:58:33 <Zakim> chimezie, you wanted to respond to LeeF 14:58:47 <ivanh> chimezie: i think there is a confusion what this suggests and what bijan suggested 14:59:06 <ivanh> ... the only thing is which particular entailment regime you use 14:59:36 <ivanh> LeeF: my problem was the maturity of the different entailement regimes 14:59:56 <ivanh> ... the interplay with the query language, how would you put it into the query language 15:00:08 <LeeF> s/was the maturity/was not the maturity 15:00:10 <LeeF> zakim, mute Orri 15:00:10 <Zakim> Orri should now be muted 15:00:24 <ivanh> (scribe had to give up) 15:01:00 <ivanh> chimezie: if you separate the inclusion of ent regime from the answer itself, 15:01:11 <ivanh> ... we need to standardize this 15:01:27 <ivanh> ... otherwise we cannot move one query from one place to the other 15:01:32 <ivanh> q+ 15:01:41 <LeeF> ack dnewman2 15:01:43 <LeeF> ack dnewman 15:02:16 <ivanh> dnewman2: end user prospective, we are talking about using sparql to dynamically trigger inferencing at the point of the query 15:02:22 <ivanh> ... that to me is a very useful feature 15:02:35 <chimezie> Seems like there is 1) Which entailment regime should answers be conditioned on 2) Is this specified in the query langauge and/or the protocol 3) how are datasets composed 15:02:38 <ivanh> ... we have come across this request, and it is very high on our wish list 15:02:44 <chimezie> zakim, mute me 15:02:44 <Zakim> chimezie should now be muted 15:02:45 <ivanh> q- 15:02:57 <ericP> dnewman2, would being able to choose between two endpoints which offered no entailment and entailmentX suffice for your needs? 15:03:14 <ivanh> q+ 15:03:29 <chimezie> I don't think SPARQL-DL requires (USING ... for example) 15:03:47 <SteveH> what about using entailments in some parts of the query and not others for example. is an example of something that curretn systems do, but this proposal does not address 15:03:48 <chimezie> it just expects additional answers to queries WRT OWL-DL entailment 15:04:12 <chimezie> SPARQL-DL being Bijan's proposal 15:04:48 <LeeF> ivanh: Bijan's proposal [SPARQL/OWL] describes what answers to expect when querying under DL entailment 15:05:04 <LeeF> ... there are other entailment regimes 15:05:07 <ericP> foaf-smushing is a high-profile candidate entailment 15:05:11 <LeeF> ... no answer for how to choose it 15:05:20 <chimezie> SteveH: I'm not sure how you can specify the conditions on a query at such a level of granularity (i.e., specific parts of the query) 15:05:20 <LeeF> ... one possibility is multiple query endpoints 15:05:31 <LeeF> s/SteveH:/SteveH,/ 15:05:45 <SteveH> chimezie, well, jena does that now I believe, and my old systems did too 15:06:01 <SteveH> chimezie, it's not really that hard, and quite essential 15:06:04 <SimonS> we do multiple graphs instead of multiple endpoints. 15:06:13 <SteveH> ditto 15:07:30 <AndyS> ditto 15:08:07 <SimonS> Views can then be used to combine the results from multiple graphs. 15:08:22 <SteveH> ivanh, the proposal where you give an entailment regime for the whole query is sufficiently behind the state of the art that I don't believe it covers the common use-cases 15:08:53 <ivanh> SteveH, let us discuss this on the list. I may not understand the issue then 15:11:28 <chimezie> I guess I'm more concerned about how the user specifies exactly the semantics of 'partial' entailments like that, not so much how they are implemented <LeeF> Adjourned. # SPECIAL MARKER FOR CHATSYNC. DO NOT EDIT THIS LINE OR BELOW. SRCLINESUSED=00000352