Provenance Working Group Teleconference

08 Dec 2011

See also: IRC log


Yogesh, pgroth, Luc, adamretter, sandro, [IPcaller], dgarijo, [ISI], stain, +1.706.461.aabb, MacTed
Paul Groth
Yogesh Simmhan


<trackbot> Date: 08 December 2011

<pgroth> Scribe: Yogesh Simmhan

<pgroth> it's all setup yogesh

thanks Paul

<Luc> Hi Adam

<Luc> Did you see my message last week? Shall we try to meet in London before Xmas?

<adamretter> Hi Luc, no i think I must have missed your message - let me check..

<adamretter> Luc: I cant see anything, which addr did you use? @exist-db.org ?

<pgroth> http://www.w3.org/2011/prov/meeting/2011-12-01


<pgroth> PROPOSED to accept the minutes of the Dec. 1 telecon

<dgarijo> +0 (I wasn't here last week)


<smiles> +1

Paul: more votes for the minutes?

<dgarijo> Satya:+1

<dgarijo> Stian:+1

<pgroth> Accepted: Minutes of Dec 1, 2011 telecon

<pgroth> http://www.w3.org/2011/prov/track/actions/open

Paul: Action item on F2F meeting, holidays

<tlebo> I just closed the last of several issues in my ACTION: https://www.w3.org/2011/prov/track/actions/49

Paul: Satya, Yolanda were to close open items

Yolanda: They are closed now

Paul: Closing all action items

<pgroth> http://www.w3.org/2011/prov/wiki/F2F2

Paul: Reminder about F2F. Hotels listed. Contact Paul is you have any questions. Book hotels early.
... Paul working on video for conference call at F2F.

<tlebo> +q

tlebo: Should people in boston get together rather than everyone connect separately?

Paul: Asked in poll. Send email to mailing list if there is interest. Did not seem so.


Paul: What is status of FPWD?

smiles: Copy of current doc is in repoitory.

<MacTed> it would help me (and I imagine others) to have the agenda include links to docs under consideration ... I always lose time tracking them down (and am never quite sure I'm looking at the right stuff)

<pgroth> sorry MacTed

<pgroth> I'll remember next time

YolandaGil: Will be useful to include diagram present in prov-dm document

<stainMobile> +1 on diagram inclusion

<pgroth> Primer: http://dvcs.w3.org/hg/prov/raw-file/default/primer/Primer.html

<MacTed> thanks, pgroth

YolandaGil: Remove section on complimentarity. Note that WG is looking into it.

smiles: There was a vote last week that there was not going to be changes in doc.

<Luc> we can vote again

smiles: What is the group's suggestion on making changes or not at this point?

Luc: We can vote again next week if we edit it.

<stainMobile> there will be few on next weeek call

Luc: Work on asusmption that vote will be positive next week. Make the changes and we can take a vote next week.

<Luc> +1 to Paul's suggestion

Paul: Send mail so that people who are not around next week can review the changes and vote next week


Luc: Suggest Paul/Luc notify the group that there will be a request for vote by email for the changes

Paul: Ready to go. Wanted feedback from W3C (sandro, evan) on the abstract. It seemed terse. Just got feedback. Just need to send email


Paul: Releasing PROV-O as FPWD

Satya: Scheduled to be released on Tue, Dec/13

<stainMobile> i have done the directory

Luc: Satya needs to get everything ready and respond to Dennis so he can do the checks.

Satya: Will be using sub-folder than a separate branch. Can make changes tomorrow.

<dgarijo> we're also working on the best practices document

Paul: FPWD are coming out before christmas. We need to send announcement again after new year to get feedback.

<stainMobile> ontology/fpwd/ in dvcs

Paul: sandro, what is automatic process in W3C?

<stainMobile> @dgarijo, yes, need a schedule for BP !

Sandro: It will appear in W3C main page. We can also send to other W3C community mailing lists. We can go ahead and publicize as soon as it is published on w3.org frontpage. We can send a link to that
... It may be automatically tweeted

Luc: Are we proposing to release best practise doc as FPWD?
... There was no vote on it since doc did not exist.
... Need to decide if we should do an internal review and vote to release it.

Paul: We have to properly circulate it

<stainMobile> do we need it fpwd at same time or refer to live version?

<stainMobile> sorry, Im on bus.. not best connection

dgarijo: Need more work. Only 3 people on the call, so we did not have consensus.
... Will meet next monday
... The current doc is linked from PROV-O

Paul: Is there a need to rush it to FPWD?

dgarijo: We can release it later.

<stainMobile> just ED in dvcs

<stainMobile> I would prefer to not rush it

<dgarijo> @Luc: +1

Luc: The authors should be convinced that the doc is ready to be released internally. We should then read and decide on it. Not there yet.

prov-dm issues

Paul: Several issues raised. Editors want to discuss.

Luc: Need to decide if we are ready to release the second working draft.

<pgroth> PROPOSED: Release PROV-DM as a second first public working draft

<pgroth> PROPOSED: Release PROV-DM as second public working draft

<sandro> Like "WD2"

<smiles> +1

<tlebo> +1

<dgarijo> +1

<MacTed> +1

<jcheney> +1

<stainMobile> 0 (University of Manchester) -- not had chance to review it yet

<sandro> +1


Satya: +1

<pgroth> ACCEPTED: Release PROV-DM as second public working draft

Luc: Many issues to discuss. Decided to try and address several points in the WD3.
... To give a better defintion of recipe.
... Issues raised by satya and others has been addressed by email.

<Luc> http://www.w3.org/2011/prov/wiki/ComplementarityUseCases#A_proposal

Please review them and respond by email.

scribe: Thanks for submitting use cases to the Wiki.
... Paolo has made proposal to define "asymmetric viewOf and a symmetric complementOf with the entailment"

Luc: would like a sense of what the WG thinks

<jcheney> How do the definitions differ?

satya: will need some time to read thru that and comment by emails. Needs to leave now.

<stainMobile> name "complementOf" must go, another big issue is the undefined "overlapping characterisation intrval"

<tlebo> just looking at definition now: "e1 viewOf e2" means that the attributes of e1 are a subset of those of e2 --- seems backwards. Wouldn't e1 have MORE attributes? We're being more specific/contextual by adding more details.

<adamretter> tlebo: I did wonder that - but after a while it makes sense to me

<stainMobile> viewOf(a,b) should require that the attribs of B was always true for A, and As attrib true during B

<GK1> Jumping in late - the asymmetric "viewOf" seems overcomplex - it seesms to me that the "inrtersectionof validity interval" is redundant, as I'd expect the validty interval of e1 to be a subset of the validity interfal of e2.

(finding it hard to hear jcheney)

<jcheney> typing because behind slow internet connection

<GK1> (Sorry for late arrival - had some home commitments.)

<jcheney> just asking for clarification of what the proposed new definitions are

<tlebo> (BTW, I finally see how symmetric and asymmetric are being used. - that definition helped)

<stainMobile> @GK1 exactly, e2 must time:intervalContain e1

Luc: GK1, can you explain what you typed on IRC?

<stainMobile> gk, sound?

<tlebo> @adamretter, what made you accept the swap of subset?

<stainMobile> no sound from gk

<GK1> Have no audio yet ... still firing up other computer.

<GK1> ... @stian, you seem to agree - can you explain?

Luc: have a question about using attribute in this definition

<jcheney> My question is answerewd by the wiki page, will read and comment on it.

<tlebo> +1 to concerns about using attributes to define viewOf.

Luc: FPWD had a strong notion of attribute

<adamretter> tlebo: e.g. - "Luc in Boston" viewOf "Luc" - so Luc in boston is a more specific version of Luc so it is a view of luc, but only when he is in boston - thats how i understood thatr

Luc: It made sense to define WasComplementOf based on attributes.
... Now, attributes may not even characterize entities.
... Should we define based on number or inclusion of attributes?

(finding it hard to hear stainMobile)

<stainMobile> sorry, bus noise..

Paul: @stainMobile said it is more hierarchical, and attributes are not as important as they once were
... We need an imprecise view of wasComplementOf. Current definiton is very precise about subsetting.

<tlebo> I have been thinking of an imprecise viewOf much like skos:broader.

Paul: We should be able to express viewOf without much semantics.

<tlebo> stephen: a viewOf b, then a is a time interval that B is in.

<GK1> I think a key feature of e1 viewOf e2 means that any (non-account-scoped) provenance assertions about e2 are also true of e1.

<stainMobile> viewOf: hierarchical, not bound to attribute, just a way to say that e1 descrived what e2 described, and e2 contains fully e1 timespan

<Luc> yes, stephen, your interval inclusion is not captured in this definition

<tlebo> simple temporal containment (and avoiding attribute discussions)?

<stainMobile> not just partial overlap as in wCO has now, that is not aa useful

stephen: We cant define intervals in terms of attributes but time scales and identity

<tlebo> +1 to using only temporal containment for viewOf and avoiding attributes.

<GK1> I think temporal containment applies in most practical cases I can think of ... and that may be the simple way to proceed ... but I'm not sure if we might find a different way of looking at this that does not depend o time interval nesting.

Luc: Stephen, did you put a usecase in the wiki page?

Stephen: yes, in th last hour

<jcheney> I will try to take theis issue into account in semantics draft

<stainMobile> I am going to do a seperate proposal about time relation

<tlebo> http://www.w3.org/2011/prov/wiki/ComplementarityUseCases#State_changes_and_relating_entities_at_different_timescales_.28stephen.29 was straightforward

Stephen: One enough details of the attributes to help state both are the same entity.

<stainMobile> it must be same thing, and contained time

Luc: is that not what SameAs does in OWL?

<stainMobile> that is what viewOf should be

<GK1> Not the same as "sameAs" (sic)

Stephen: But that is stronger.

<stainMobile> which would allow different prov statement

Luc: is GK1's view the same as Stephen?

GK1: For practical purposes, yes.
... Was trying to think of e.g. that has value over a spatial field wth different types of containement
... It does seem like an interval containment

Paul: Stephen commented that there has to be a sub view of something else.

<GK1> +1 paul comment about sub-view

Paul: Can Luc in soton be a sub view of Luc in his entire life?

Stephen: Yes
... There is a hierachical nature.

<stainMobile> +1 for subview

<GK1> Luc today <= luc in soton <= luc in his lifetime

<smiles> would have to be contiguous period of time in southampton

Stephen: And Luc@life is not a view of Luc@soton

<stainMobile> not sure wr need complement of?

Stephen: Anti-symmetric

<GK1> ... unpess a==b

<smiles> not just luc in southampton ever (as he will travel in and out)

<stainMobile> can contain eachother only if same interval

Luc: is anyone against interval containment?

<smiles> yes, supportive

Luc: Should it be the only notion?

<GK1> I don't mind about something else.

<tlebo> supportive of just temporal intervals, until we see it written up.

Luc: Do we retain wasComplementOf?

<stainMobile> +1 to interval containment, -1 to wasCompOf

Paul: Looking for a lighter view in addition to interval containment

<stainMobile> its like a sibbling view with time overlap

<tlebo> I think wasComplementOf is a derivative, ancillary relation that is based on a viewOf "hierarchy"

<GK1> @paul - that's kind of "top" of the interval space, isn't it?

Luc: For both attributes and intervals, we may not be able to verify

<GK1> @luc, think that's OK - it becomnes an existential assertion/.

<StephenCresswell> +q

<stainMobile> and specially if not yet finished..

<GK1> @luc; I think this relates to the idea that attributes are most important foir interop with other systems.

<stainMobile> what if we split it in two, viewOf just needs time overlap

Stephen: If we cannot say anything about intervals, two weak notions are (1) complementOf, where the intervals overlap, (2) or both are views of a wider entity

<stainMobile> and state time contain as welk seperateky

<GK1> @stephen +1

Stephen: We still have viewOf, but it is very long term spanning all time

<stainMobile> +1

smiles: if entity exists, it cannot be verified in itself. Why does ti matter if the asseriton can be verified?

<tlebo> smiles: we can't verify that an assertion that a document exists is true, so verification can't be necessary.

<GK1> @smiles +1

Luc: acknowledges the point

<GK1> (to repeat...) I think a key feature of e1 viewOf e2 means that any (non-account-scoped) provenance assertions about e2 are also true of e1.

<GK1> My vote would be -0 - i.e. prefer not talking of attributes, but could live with it.

<stainMobile> -0 as well

Stephen: Only interval containment of the same entity.

<smiles> i don't see a need for referring to attributes, just say "talking about same thing"

<tlebo> @GK1, no, the assertions on e2 apply differently to the assertions on e1 - you can't just copy/paste them up the viewOf hierarchy.

<stainMobile> (The discussion made me get off bus 2 stops late!)

Luc: will make a proposal to review next week

<Luc> http://www.w3.org/2011/prov/wiki/RecipeDiscussion

<GK1> a viewOf b => exists(realworldobjec) s.t. a and b are each viewOf realworldobject

Luc: talking about recipe. See link.

<tlebo> (@GK1, l like FRBR and "maker" - maker of the paperbound is not the maker of the story it conveys)

Luc: Use the term Plan rather recipe?
... It seems that Plan can evolve, so should a Plan be a kind of entity?

<stainMobile> plan fits better than recipe with activity

Luc: If so, then do we have a new relationship hadPlan: activity -> entity?

<GK1> @tlebo - I worry about trying to make to strong a parallel with FRBR.

Luc: Is this a specialization of wasAssociatedWith? If not, how do we distinguish this new relationship from wasAssociatedWith?

<tlebo> @GK1, we won't mention FRBR, just like we don't mention reification for QualifiedInvolvements :-)

Paul: Plan to to be associated with activity. So a specialization.

<tlebo> is "wasAssociatedWith" the too-weak relation between agent and activity?

<tlebo> (too weak in name)

<stainMobile> plan is not used by activity, but might be usrd by agent who is also activitu

<stainMobile> activity

YolandaGil: the plan may be "The Plan" in some cases. But in most cases, it may be more than one and may evolve

<stainMobile> have to go, sorry

YolandaGil: Maybe the lesser the commitment we make to the plan, the better?

<tlebo> yolandaGil: the plan may not apply universally to an activity, there may be multiple plans at different times (used by different agents).

YolandaGil: What if there are multiple plans depending on how far before the activity it is defined?

<Luc> hadPlan: 1 to many relationship?

YolandaGil: Activity points to one or more plans.
... Someone can make the plan an entity and make further associations.

<YolandaGil> +1 for plan

Paul: Always though Plan was going to be a hook to other things, but not to define a planning language
... suggest we dont have cardinality or semantics to that.

<GK1> @pgroth +1 (we just had similar discussions w.r.t. our project :)

Luc: Paolo and Luc will write this up and circulate it.

<tlebo> @paul, but does the hook apply to JUST the activity (universally), or does the hook apply to the controlling agents' control of the activity?

<YolandaGil> Your proposal for plan sounds good Luc

<Luc> wasAssociatedWith

<tlebo> -1 naming, no counterproposal :-(

Luc: Any counter proposal for WasAssociatedWith?

<YolandaGil> I like it

<GK1> @tlebo I think the hook may apply directly to something like the activity (only), but that in turn may be linked to to other aspects.

Luc: In the absence of counter proposal, we cant resolve. We can wait for vote after christmas.

<pgroth> trackbot, end telcon

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.136 (CVS log)
$Date: 2011/12/08 17:04:26 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.136  of Date: 2011/05/12 12:01:43  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/th/the/
Succeeded: s/ti/it/
Succeeded: s/at doc/a doc/
Succeeded: s/I like/Like/
No ScribeNick specified.  Guessing ScribeNick: Yogesh
Found Scribe: Yogesh Simmhan
Default Present: Yogesh, pgroth, Luc, adamretter, sandro, [IPcaller], dgarijo, [ISI], stain, +1.706.461.aabb, MacTed
Present: Yogesh pgroth Luc adamretter sandro [IPcaller] dgarijo [ISI] stain +1.706.461.aabb MacTed
Found Date: 08 Dec 2011
Guessing minutes URL: http://www.w3.org/2011/12/08-prov-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]