12:41:55 RRSAgent has joined #prov 12:41:55 logging to http://www.w3.org/2011/07/07-prov-irc 12:41:57 RRSAgent, make logs world 12:41:57 Zakim has joined #prov 12:41:59 Zakim, this will be 12:41:59 I don't understand 'this will be', trackbot 12:42:00 Meeting: Provenance Working Group Teleconference 12:42:00 Date: 07 July 2011 12:42:13 Zakim, this will be PROV 12:42:13 ok, Luc; I see SW_(PROV1)8:00AM scheduled to start 42 minutes ago 12:42:40 Chair: Paul Groth 12:42:59 Agenda: http://www.w3.org/2011/prov/wiki/Meetings:F2F1Timetable 12:43:23 rrsagent, make logs public 12:43:45 Scribe: Paolo Missier 12:49:18 TOPIC: Session 5: PAQ TF 12:56:08 GK has joined #prov 12:56:59 zednik has joined #prov 12:58:29 pgroth has joined #prov 12:58:49 SW_(PROV1)8:00AM has now started 12:58:55 +Meeting_Room 13:00:12 khalidbelhajjame has joined #prov 13:01:44 Paolo has joined #prov 13:01:56 RRSAgent, pointer? 13:01:56 See http://www.w3.org/2011/07/07-prov-irc#T13-01-56 13:02:04 rrsagent, make logs public 13:02:46 ericstephan has joined #prov 13:02:55 meeting: Prov F2F1 Day 2 13:03:17 sandro has changed the topic to: PROV F2F1 - Conference Code is DIFFERENT: 77681# (note the "1") Webcam: http://www.w3.org/People/Sandro/webcam 13:03:29 topic: PAQ TF (session 5) 13:04:17 +zednik 13:04:24 tlebo has joined #prov 13:04:25 Deborah has joined #prov 13:04:34 jcheney has joined #prov 13:05:15 SamCoppens has joined #prov 13:06:21 StephenCresswell has joined #prov 13:06:37 smiles has joined #prov 13:06:40 Scribe: Paolo 13:06:41 http://www.w3.org/2011/prov/wiki/images/1/1d/Paqtf_status.pptx 13:07:01 Yogesh has joined #prov 13:07:03 Vinh has joined #prov 13:07:24 JimMcCusker has joined #prov 13:07:26 going through the slides 13:07:28 slide 1 13:08:05 GK2 has joined #prov 13:08:38 Paulo has joined #prov 13:08:42 slide 2 13:09:39 Can someone re-post the link to the slides? 13:09:41 3 13:09:50 +??P1 13:09:51 http://www.w3.org/2011/prov/wiki/images/1/1d/Paqtf_status.pptx 13:09:52 http://www.w3.org/2011/prov/wiki/images/1/1d/Paqtf_status.pptx 13:10:02 zakim, ??p1 is me 13:10:02 +GK; got it 13:10:12 thanks 13:10:31 4 13:10:43 RyanGolden has joined #prov 13:11:01 slide 5 13:11:13 can you post the URL to the slides again? 13:11:25 Satya has joined #prov 13:11:35 http://www.w3.org/2011/prov/wiki/images/1/1d/Paqtf_status.pptx 13:12:07 slide 6 13:13:47 pausing for the projector... 13:13:51 IlkayAltintas has joined #prov 13:14:22 (third option is proposal from Luc) 13:14:28 on this slide 13:14:38 +olaf 13:14:48 4 was a proposal from GK 13:15:06 bullet 5 was proposed by Stian 13:15:15 olaf has joined #prov 13:15:50 so slide 6 summarize proposals on first question 13:16:07 that was "Given information regarding where to access data on the provenance of a resource state representation, what form does that information take and how do we obtain the provenance data? " 13:16:41 slide 7 13:17:05 proposals for Q2, embedding provenance into an HTML doc: "How can a browser find the information on where to access provenance data, referred to above, for an HTML document that was downloaded, so that its provenance may be retrieved? " 13:17:24 bullet entry 1 from GK 13:17:31 bullet 2 from Luc 13:18:02 3 also proposed by Luc 13:18:21 is that the slideset Simon sent on Tuesday? 13:18:30 no it's from today 13:18:30 bullet 4: from Khalid 13:18:36 can someone paste the url again 13:18:50 @olaf: this is the set: http://www.w3.org/2011/prov/wiki/images/1/1d/Paqtf_status.pptx 13:19:05 thanks! 13:19:06 slide 8 13:20:04 That's pretty close to what I meant 13:20:11 GK: remember KISS :-) 13:20:25 :) 13:20:29 q? 13:21:15 slide 9: hopefully uncontroversial suggestions for decisions 13:21:27 q? 13:21:37 q? 13:21:50 "There may be data regarding the provenance of a thing accessible from multiple sources." 13:22:18 item 1 -- there may be multiple services providing provenance, or multiple prov URIs for an entity state 13:22:39 this has implications for access 13:23:12 smiles: not the intent to suggest the data is the same. 13:23:23 Different provenance from different sources could be different, even inconsistent. IMO. 13:24:00 Proposal: Provenance of a thing can be found at multiple sources 13:24:13 +1 to "There may be data regarding the provenance of a thing accessible from multiple sources." (and this provenance information may differ) 13:27:58 Tim: can we just state provenance without referring "of ...(Bob etc)"? 13:28:30 Tim: main point is multiplicity 13:28:53 I would say that (provenance data) is a web *resource* - the state representation is what is actually transferred. 13:29:04 provenance metadata may be available from many sources and need not be globaly consistent? 13:29:29 Proposal: Provenance of a "bob" can be found at multiple sources 13:29:36 +1 13:29:45 +1 13:29:46 +1 13:29:49 +1 13:29:49 +1 13:29:50 +1 13:29:50 +1 13:29:50 +1 13:29:52 +1 13:29:54 +1 13:29:54 +1 13:29:56 +1 13:29:58 +1 13:30:02 +1 13:30:15 +1 13:30:22 -1 : should be "provenance of EntityState" OR we remove terms from the model. 13:30:27 I redact the use of consistent in the earlier comment 13:30:44 this is about location. the point of consistency to be addressed later 13:31:22 I retract (we are not talking about the model) 13:31:31 0 13:31:44 is this source in the "real world" or it does not matter? 13:32:10 accepted: Provenance of a "bob" can be found at multiple sources 13:32:44 Smiles: we are not discussing nature of source just now 13:33:05 Anyone can say anything about anything... including provenance. 13:33:25 +1 to AAAP 13:34:43 q? 13:34:54 KISS: start with easy cases, then address identified gaps. 13:35:31 Paulo: whatever the solution to provenance encoding, it should be non intrusive wrt the underlying data 13:35:49 paulo: not changing a bit of a BOB while still being able express provenance of a BOB. 13:36:09 smiles: we are agnostic, some encodings may be intrusive 13:36:26 Paulo: at least one encoding should not be intrusive 13:36:39 smiles: the opposite may also be true 13:36:54 Proposed: The information required to obtain access to some provenance of a thing may be supplied in many different ways, and we do not aim to enumerate them all. 13:36:59 +1 13:37:01 +1 13:37:03 +1 13:37:06 +1 13:37:07 +1 13:37:09 +1 13:37:11 +1 13:37:13 +1 13:37:16 +1 13:37:16 +1 13:37:17 +1 13:37:18 +1 13:37:23 +1 13:37:28 +1 13:37:29 +1 13:37:38 +1 13:37:59 ACCEPTED: The information required to obtain access to some provenance of a thing may be supplied in many different ways, and we do not aim to enumerate them all. 13:39:49 Ilkay: change "thing" in previously accepted point 13:40:00 q+ 13:40:15 q+ protocols that allow third parties to submit pointers to provenance of a BOB. 13:40:24 PROPOSED: The WG effort will concern how the provider of a BOB can supply information required to obtain access to some provenance of that BOB (which may, as a side effect, include recommendations on how others can do the same). 13:41:10 on third point 13:41:40 ACCEPTED: The information required to obtain access to some provenance of a BOB may be supplied in many different ways, and we do not aim to enumerate them all. 13:42:01 olaf: this is focused on apporaches to provide provenance, but it is also important to allow provider to associate provenance /as part of the "thing"/ 13:43:05 pgroth: propose to replace supply with obtain 13:43:16 luc: or "embed"? 13:44:31 pgroth: first-party ONLY providing access. OUT OF SCOPE: third party providing information about accessing provenance. 13:44:38 pgroth: example: WG is concerned with how a /data provider/ supplies provenance about it 13:44:48 q? 13:44:49 q? 13:45:16 q+ 13:45:22 q+ to ask about third parties submitting pointers to first parties (which first parties can choose to include in their access descriptions) 13:45:28 smiles: alternative is not to restrict on who supplies provenance, but that's too broad 13:46:01 ack loaf 13:46:07 ack ola 13:46:08 q- 13:46:26 (e.g. First Party is New York Times that makes article) 13:46:44 satya: do we need a distinction between first party and third parties (as in ex. above) 13:47:39 are we failing to support "down stream" provenance of a BOB created? 13:48:04 satya: eg embedded HTML link may not be from first party -- distinction may be difficult to make in practice 13:48:17 luc: provider vs author. 13:48:19 Third party provenance: NYT vs blogger not distinguished - trust is an orthogonal issue. The resource provider has privileged access for indicating provenance sources - I think that's unavoidable. 13:48:26 q? 13:48:27 satya: "first party" gets blurry 13:48:42 q? 13:48:42 ack Satya 13:48:42 q- 13:48:44 satya: need to clarify "author", "first party", third party, "provider" as they may get blurred 13:48:48 ack tlebo 13:48:48 tlebo, you wanted to ask about third parties submitting pointers to first parties (which first parties can choose to include in their access descriptions) 13:49:06 q+ 13:49:31 tlebo: WG effort seems to be focused on first parties -- does this allow us to accept third party provenance of Bob's? 13:49:34 The point of the scope is to simplify things: so initially, make choices that simplify. Later consider alternatives. 13:50:20 q+ 13:50:29 tlebo: a provider should be able to accept additional provenance of its Bob from a third party. this is key to establish a web of provenance 13:50:46 It's not up to the provider to "accept" third party provenance assertions. The web allows anybody to say... 13:50:49 How is defining the scope of the user affecting the workload of the PAQTF? 13:51:06 ack Luc 13:51:22 It avoids having to initially consider problems of third party discovery. 13:51:30 ^^ @satya 13:51:54 @GK How? What are the problems of third party discovery? 13:52:09 q? 13:52:15 ack Yog 13:52:22 smiles: first party provenance-of-BOB publishers accepting third party provenance-of-BOBs and choosing to include them is NOT out of scope. 13:52:24 q- 13:52:38 q+ 13:52:40 @Satya: I have a resource without provenance. You publish provenance about that resource. How does someone else discover that provenance? 13:52:58 Luc, smiles: w elmit the scope for the first draft only 13:53:15 s/w elmit/we limit 13:53:40 Sure, first draft only. 13:53:56 @GK They either query the resource itslef (for embedded link) or they "llok it up on the web" 13:54:08 itslef> itself 13:54:09 "letting third parties do what they want" is insufficient, because we are failing to support DISCOVERABILITY (which is part of ACCESS). 13:54:30 @Satya: first approach requires provider to cooperate. Latter is default - what more to specify? 13:54:38 q? 13:55:05 yogesh: starting point for discovering provenance is the provider 13:55:19 ack tl 13:55:45 @GK why as a WG we are mandating "bob" providers to give access to the provenance of "bob"? 13:56:32 @Satya, we're not. Just focusing first on those that want to. 13:56:44 tlebo: NYT should be able to supply provenance of its own image, but also of new versions of that image that may have been manipulated by somebody else 13:56:50 q+ 13:57:28 q- 13:57:29 pgroth: the example is orthogonal to this proposal 13:57:49 Nothing is being excluded as an eventual possibility... 13:58:03 provenance being downstream etc is not the point here 13:58:08 q? 13:58:08 @GK Anybody may want to - why should we discriminate between "first", "second" etc. providers? 13:58:14 ... my view (an experience) is that when a simple solution is on the table, it's much easier to see how tio engibneer more advanced solutions. 13:58:30 @SDatya. Who said anything about disciminating. 13:59:26 Luc: should the draft editors be free to decide, as per previous point just accepted 13:59:50 @GK We are trying to categorize "bob" providers as first, second, etc parties 13:59:54 We rule out nothing in the longer term. 14:00:27 [[ Apologies, I need to step out for ~60 minutes, with my laptop, so no webcams either. Very sorry. ]] 14:00:55 @Satya - I think one must recognize that the provider of a resource has control over metadata that accompanies that provision. I see that is a given. Beyond that, not attempting to categorize. 14:01:30 q? 14:01:35 @GK ok - then we don't have to specify this explicitly since it is open/implicit anyway 14:01:38 q- 14:01:38 just moving on from here 14:01:40 Moive on? 14:01:48 slide 10 14:03:44 point 1: 3 options 14:03:49 q? 14:03:56 q+ 14:03:57 q+ 14:04:06 q? 14:04:51 I think you meant to say "How do you find what it is provenance _of_"? 14:05:00 is this for this TF? the model has a notion of "prov container" 14:05:41 smiles: question is, what do you need (I,L, etc.) to gain access to provenance 14:05:48 q? 14:06:56 @paolo - I agree it's a model issue, but I think it's highly desirable that retrieved provenance data identifies what it is referring to. 14:07:04 ack Yogesh 14:07:07 ack Luc 14:07:53 satya: is the question how you associate P and I? 14:08:45 That is an assumption. 14:09:14 satya: realistic scenario is that you don't start from provenance, you start from the data (entity state) 14:10:43 One could imagine doing a web search to provide the most accurate available instance of some some data: in this case one might find provenance first, then use that lo locate the resource. 14:10:59 clarification: I is the URI (reference to) an entity state, not the entity state itself 14:11:33 q? 14:11:44 I think we've found another Bob... 14:12:02 satya: not clear how we identify entity states in the first place 14:12:08 My assumption is that BOBs have URIs (or may have URIs) 14:12:50 For a genuinely static resource, it's possible R == I (resource URI == BOBN URI) 14:13:39 -zednik 14:13:40 GK maybe you should get in the queue? 14:14:14 q+ 14:14:19 q+ 14:14:55 zednik has joined #prov 14:15:44 q+ 14:16:18 (I'd get on the queue, but the conversation seems to keep jumping around - not sure what I really want to respond to.) 14:16:27 +zednik 14:17:00 zakim, close the queue 14:17:00 ok, pgroth, the speaker queue is closed 14:17:00 yogesh: difference b/w options a) and b) seems to be one of granularity 14:17:39 q? 14:17:40 q? 14:18:43 satya: are we assuming we have a "provenance container" with a single URI P for a set of provenance assertions? 14:19:44 My default position is that provenance is on the web, and as such may be a resource, and as such may (and often should) have a URI. That deals (IMO) with 80-90% of the access mechanism. 14:20:21 smiles: either you need the ID (I) of a specific Bob, or the association is apparent and that's not needed 14:20:47 q? 14:21:15 pgroth: we seem to be discussing the phrasing of the issue, rather than any specific solution 14:21:27 ack satya 14:21:44 i think the problem is important but not well posed 14:22:11 ack Luc 14:22:21 satya: don't think we should get into the "inverse relationship" P -> I 14:23:21 I would say that for the purpose of provenance *interchange*, it should be explicit what the provenance is about, even if it's implicit in its original form. 14:25:09 pgroth: it must be an issue, cannot just be ignored. 14:25:11 dcorsar has joined #prov 14:25:34 I can easily imagine a single RDF graph (provenance resource) that actually contains provenance of several BOBs. 14:26:01 Ah, it's not about saying the resource is about one subject *only*... 14:26:27 what slide is option B on? 14:26:31 ... just requiring that any given provenance information is explicit about what it's about. 14:26:37 slide 10 14:26:40 Edoardo has joined #prov 14:26:48 @tlebo slide 10 14:27:01 The problem with (a) is the "single" 14:27:28 q? 14:28:17 Drop the *only* 14:28:32 Yes: many-to-many 14:29:21 thanks GK :-) 14:29:42 (Actually, solutions that solve single issues cleanly often scale up very well.) 14:29:58 paolo: Bob-to-provenance is a M-M relataionship and we need a mechanism to traverse it in both directions 14:30:16 s/relataionship/relationship 14:30:48 satya: Bob->provenance is the only direction we can hope to traverse it 14:30:49 q+ 14:30:54 q- 14:31:04 sorry james 14:31:44 q- 14:32:49 (break) 14:32:56 -olaf 14:33:30 back in 15 minutes 14:34:14 @smiles: your slide 10, bullet 1, (a), suggest rephrase "(a) It is apparent from the data itself what thing(s) it describes the provenance of" 14:35:00 -zednik 14:45:05 @GK , @smiles I like that rephrase 14:45:39 +zednik 14:45:51 I have to go now; will try to tune in later again. 14:49:11 TOPIC: MODEL task force 14:49:37 SamCoppens has joined #prov 14:49:37 edsu has joined #prov 14:49:53 topic Model 14:49:54 ericstephan has joined #prov 14:50:42 Scribe: SamCoppens 14:51:01 Topic: Model 14:51:34 PROPOSED: assertions about time are useful but are optional 14:52:10 +1 14:52:12 +1 14:52:14 +1 14:52:16 +1 14:52:17 +1 14:52:17 Deborah has joined #prov 14:52:18 +1 14:52:18 +1 14:52:21 q? 14:52:23 0 14:52:25 +1 14:52:31 +1 14:52:41 StephenCresswell has joined #prov 14:52:56 Paolo_ has joined #prov 14:53:03 +1 14:53:05 Yogesh has joined #prov 14:53:10 +1 14:53:12 +1 14:53:23 +1 14:53:36 +1 14:53:49 +1 14:53:51 +1 14:53:53 +1 14:54:06 +1 14:54:21 ACCEPTED: assertions about time are useful but are optional 14:55:32 luc: Ordering of events 14:56:01 q? 14:56:47 Paolo: it is important to have an understanding of the ordering of events, but it may be left out of the provenance 14:57:09 S/paolo/Paulo 14:58:44 (Saying that a resource is used before it is generated is not the same as saying nothing about ordering.) 14:58:59 PROPOSED: separate Time from (Event) Ordering 14:59:59 Simon: what is the reason to include Ordering 14:59:59 +q 15:00:24 q+ 15:00:28 zakim, open the queue 15:00:35 ok, Luc, the speaker queue is open 15:00:42 Chair: Luc Moreau 15:00:47 q+ 15:00:47 zakim, open the queue 15:01:01 q? 15:01:07 +q 15:01:15 ok, Luc, the speaker queue is open 15:01:16 Luc: it is explanatory 15:01:45 akk Paulo 15:01:50 q? 15:02:42 Edoardo has joined #prov 15:03:18 Yes: not all metadata is provenance (but may still be useful, and provenance should not exclude non-provenance information) 15:03:24 q? 15:03:38 time dimension is the only criteria differentiating provenance and non-provenance metadata (my 2cs) 15:04:10 ack Paulo 15:04:14 q+ 15:04:21 q? 15:04:31 -q 15:04:50 q-\ 15:04:52 q- 15:05:32 Luc: will ordering of events be considered 15:05:33 q? 15:05:53 q? 15:06:09 ack smi 15:06:55 Simon: is this explanatory or a constraint 15:06:56 @Simon It needs to be a constraint 15:07:24 time issue 6 would lead us to talk about ccs, csp, temporal logics and dynamic logics among others. 15:07:36 @Satya oh, does it now? 15:09:45 @Simon ;) 15:09:51 Issue: consider ordering of event in model and semantics 15:09:51 Created ISSUE-27 - Consider ordering of event in model and semantics ; please complete additional details at http://www.w3.org/2011/prov/track/issues/27/edit . 15:10:30 q? 15:10:40 q+ 15:10:42 Agent definition 15:10:49 q+ 15:10:53 Subtopic: agent 15:10:57 q? 15:10:58 q+ 15:11:06 ack pgroth 15:11:07 q+ 15:11:42 q+ respond to Paul 15:11:46 q? 15:11:56 +1 to paul's view- agents should be able to stand alone 15:12:02 Paul: Agents just being able to exist in provenance description. Agents should be able to stand alone 15:13:10 Paul: Agents should not always be bound to process execution 15:13:13 ack Jim 15:13:44 Jim: Agent is entity that can perform activity, but they must not be bound to activity 15:14:01