Provenance Working Group Teleconference -- 03 Nov 2011

<trackbot> Date: 03 November 2011

<Luc> Scribe: Graham Klyne

<Luc> Graham, everything should be set up for you

OK, thanks.

<Luc> thanks for volunteering

Admin

<Paolo> zakim. ??P16 is me

<Luc> @sandro, did you see my message about not being able to create a poll?

Luc: intro - see agenda
... http://www.w3.org/2011/prov/wiki/Meetings:Telecon2011.11.03

<Luc> Proposed: to accept minutes of last week's teleconference

Luc: no AOB for agenda

<smiles> +1

<Yogesh> +1

0 (not present)

<Paolo> (missed it)

<Paolo> 0 (not present)

<Luc> ACCEPTED: minutes of last week's teleconference

F2F2 and F2F3

<Luc> January 23-24

<Luc> February 2-3

Luc: possible dates for F2F2 - 23, 24 Jan or 2-3 Feb

<Luc> Proposed locations: Amsterdam (preferably, since more options), Southampton

Luc: will set up a poll for expressing preference
... locations Amsterdam or Southampton
... asks about permissions for setting up poll

Sandro: should have permissions now

<Luc> June 22-23

<Luc> June 22-23, santa barbara

<Luc> July 2 to 6, Boston

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

Luc: Jun dates colocate with IPAW12 - http://www.ipaw.info/ipaw12/

<Luc> http://www.w3.org/2011/01/prov-wg-charter

Luc: Objectives for F2F2 meeting
... charter indicates should be at last call for PROV-DM and PROV-O
... but don't feel this is realistic for Jan/early Feb
... But want to identify what remains to be done to release last calls.
... For PAQ, also want to identify what needs to be done to go to last call.
... as a NOTE, not REC-track

PROV-XML (deliverable 5) .. similarly want to identify what id needed.

Luc: Primer due after M12(?), so identifying roadmap at F2F is goal
... similarly for best practice "cookbook"

<pgroth> +q

Luc: and call for implementations (?)
... thus... 8 topics is a good fit for 8 sessions at F2F
... inputs needed will be:

<scribe> ... new version of PROV-DM

UNKNOWN_SPEAKER: released version of PROV-O
... second working draft of PAQ, including querying
... first version of semantics
... first version XML serialization, primer
... outline of work plan for deliverable 6 (which is that?)

pgroth: have question about availability for next F2F in Europe

<smiles> +1 (at least for the Jan dates)

<MacTed> -1

+0.5

<Yogesh> -1

<Christine> Paul - I joined late. Could you repreat the dates please?

<zednik> -1

<satya> -1

<Paolo> Luc: possible dates for F2F2 - 23, 24 Jan or 2-3 Feb

<dgarijo> +1 ( I think)

<Christine> +1 Feb

<Paolo> Proposed locations: Amsterdam (preferably, since more options), Southampton

(that probably shouldn't be minuted)

<Paolo> +1 (so far)

<Yogesh> Will not be able to travel to Europe...Can join by phone

Luc: those with -1 response: is it dates or location?

Stephan: both

Satya: availability of dates

Luc: feedback on objectives for next F2F, and inputs requested?

jcheney: does semantics mean ontology, or separate deliverable. If separate deliverable, am I the leader? Need to figure out goals for separate semantics document.

Luc: asks paul to put agendum in for next week about semantics document

<dgarijo> yes

<smiles> +1

<satya> +1

<dgarijo> (+1)

<zednik> +1

<Luc> Proposed: to accept the objectives and inputs for F2F

<Luc> Proposed: to accept the objectives and inputs for F2F2

<Paolo> +1

<jcheney> +1

<Luc> Accepted: the objectives and inputs for F2F2

PROV-O

(See http://www.w3.org/2011/prov/wiki/F2F2)

Satya: report from PROV-O teleconference... alignment of DM and ontology, and other progress
... how to model entity roles. Two proposed approaches:
... (1) specialize ?? (subproerty of (?)
... (2) use a new class
... given complexity of information being modelled.
... felt creating n-ary class was more intuitive
... explore option of allowing both approaches

<tlebo> my impression from the OWL telecon was that we included both techniques into PROV-O, so neither was excluded.

Satya: actions of Luc, Daniel to define appropriate properties for linking the n-ary classes to the appropriate entities

s.of/on/

<dgarijo> me too! :D

Luc: wasn't present for action given to me... surprised!

<tlebo> (prov:used to point directly, versus prov:usedUsage points indirectly)

Satya: we felt you would be in better position as author of OPMO technique

<dgarijo> I thought we were going to use those names already.

Luc: feels that Tim, Stephan(?) are better placed to choose names.

<tlebo> +1, I can take a stab at the names.

<zednik> I agree with tim and daniel, keep prov:used and prov:hadUsage

<pgroth> +q

Luc: propose use names from example discussed on Monday, then discuss

<zednik> +1 I am willing to help

<dgarijo> @zednik: I think he is referring to the: entityUsed (or usedCause) properties.

<satya> http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology_Meeting_2011-10-31#Meeting_Information

<scribe> ACTION: tlebo to propose names to use [recorded in http://www.w3.org/2011/11/03-prov-minutes.html#action01]

<trackbot> Created ACTION-41 - Propose names to use [on Timothy Lebo - due 2011-11-10].

<dgarijo> sure

<dgarijo> I can help

Satya: in other sections...

<tlebo> Should we shoot to have the names by Monday noon?

Satya: improving readability
... also keeping doc aligned with PROV-DM

<Luc> @tlebo, as soon as possible, yes

Satya: inference rules may change
... hopefully in next couple of days

Luc: action 40 still in progress
... http://www.w3.org/2011/prov/track/actions/40

Satya: work in progress

<dgarijo> +q

<pgroth> +q

Satya: would like to leave examples showing both approaches,

<tlebo> zeknik's proposal includes the "EntityInRole" approach, no?

<zednik> @tlebo no

??: could be confusing, maybe put examples in another document?

Satya: will introduce new sections, keep issues separated

<dgarijo> @Graham: ?? is me

<tlebo> @zednik, s/prov:hadUsage/prov:qualifiedUse/ for the qualified version of prov:used ?

<dgarijo> @Paul: yes, I was wondering exactly about the same thing. If we release 2 approaches, interoperability is going to be an issue!

pgroth: I think we need to pick one: (1) for interoperability, (2) avoinding confusion, keeping things easy to explain
... two ways of modelling same thing shouldn't happen.

<smiles> @pgroth agreed

<dgarijo> @paul:+1

<pgroth> +q

<dgarijo> @tlebo: no, entityInRole is the concept we decided to drop on monday.

<dgarijo> @tlebo: and replace it with Usage, Generation, Control.

<Paolo> +q

satya: commenting on modelling alternatives - given preferebnce for n-ary approach, don't see how this can be avoided.

<tlebo> we're just renaming the same design pattern, and including the ability to state it directly AND indirectly.

luc: I don't think either approach introduces more instances

<tlebo> +1 GK

<Paolo> @Satya: I tried to make it clear that my comment was not putting one approach against the other -- it was rather provocatively on the OWL/RDF mapping in general

<dgarijo> @tlebo: what we are doing is stoping EntityInRole from being a subclass of Entity.

luc: my objection was to introducing (classes) not in PROV-DM

<Paolo> +q

<pgroth> queue

<Zakim> GK, you wanted to ask about taking on board Ivan's comments about complexity

satya: can you show how extra nodes don't need to be added for OPMO

tim: don't see this as a contention, but new design that subsume's. Can use either direct or indirect approach

<Zakim> tlebo, you wanted to say that EntityInRole is part of the new design proposal (it includes it's more direct alternative, too).

pgroth: primer people are saying entityinrole is difficult to describe. We need one way to model.
... would like to see this written down

satya: I am not clear about several issues relating to the alternative approach

pgrioth: stephan and tim have clear idea about what needs to be done

<dgarijo> +1 to ultimate approach

satya: will not have two approaches, just the alternative approach

<pgroth> but that's rdf :-)

<Paolo> The EntityInRole approach has 19 nodes, about as many properties, and 3 new OWL classes

<tlebo> (the new "Usage, Generation, Control" qualifications of prov:used are subclasses of EntityInRole, but we won't be naming EntityInRole or asserting the subclass axioms.)

paolo: node count shouldn't be used as argument for one approach over the other (?)

Luc: one approach or two?

Satya: one approach. New proposed approach, to be included when details are clear

Luc: when will PROV-O document be finalized?

<tlebo> regarding EntityInRole: we are just renaming it AND permitting "more direct" ways to state less qualified usages.

Luc: i.e. available for WG review

<zednik> @tlebo, but the semantics of Usage are different than the semantics of EntityInRole were

<dgarijo> @tim: but we are not just renaming it. EntityInRole is no longer a subclass of an Entity!

Satya: modelling changes will take time to work through
... new editorial workover in a couple of days. Modelling changes will take longer.

<dgarijo> @tlebo: if not, then the issues with double usages and new entity creation for the uses are still there.

<tlebo> @dgarijo, I'm indifferent to whether EntityInRole is a subclass of Entity, it's fine to relax it.

<zednik> I think there is less agreement than was earlier assumed

Luc: worried that rest of author team doesn't seem to agree with new approach

<jcheney> +q

Satya: can we have another separate call today or tomorrow to work this out.

Luc: we need you to reach a decision and give the rest of us something to review

Satya: will need some time to come up with missing parts not in OPMO/OWL

<dgarijo> @telbo: you are saying that EntityInRole being a subclass on an Entity is indiferent. It changes everything.

jc: idea was raised as a thought experiment

<tlebo> @dgarijo, stop scaring @luc :-)

jc: subgroup doesn't necessarily agree, but if we can forge consensus that one approach is good enough to put out for review and comment, this would be a way forward

<dgarijo> @tlebo: ok, but I just thought we had an agreement after the other day's telecon. I'm ok about having a telecon afterwards though.

tlebo: not so much disagreeing, just ... converging on new direction?

<tlebo> apologies for making it look like we are disagreeing.

Luc: team will continue work on document, make it available to WG ASAP

<Zakim> tlebo, you wanted to say we aren't disagreeing too much.

paq document

pgoth: about same as couple of weeks ago. Have been waiting on things to settle in data model. Also in ontology - as these affect queries and handling of context
... expecting it will take about a week to bring this up to date
... TODO
... context URIs and entities
... provenance services

<tlebo> (It's been a while since I've dug through OPMO - where's the best summary?)

pgoth: incremental updates
... editorial

<satya> @Tim: http://openprovenance.org/model/opmo

pgoth: w.r.t. incremental updates: How to get provenance information incrementally? Explaining fow to deal with large amounts of provenance?

<tlebo> @satya thanks!

Luc: do you mean this will be ready by next week?

Paul: I think we can manage that

PROV-DM

<scribe> ACTION: Pau to drive new release of PAQ to WG for next week [recorded in http://www.w3.org/2011/11/03-prov-minutes.html#action02]

<trackbot> Sorry, couldn't find user - Pau

<scribe> ACTION: pgroth to drive new release of PAQ to WG for next week [recorded in http://www.w3.org/2011/11/03-prov-minutes.html#action03]

<trackbot> Created ACTION-42 - Drive new release of PAQ to WG for next week [on Paul Groth - due 2011-11-10].

<Luc> Accepted: Rename 'Entity Expression' into 'Entity Record'; similarly, rename 'XXX Expression' into 'XXX Record'.

<Luc> Accepted: Attributes are a necessary part of prov-dm. Attribute-value pairs can be optionally included in Entity Expressions and Activity Expressions.

<dgarijo> The definition from the incubator: Provenance of a resource is a record that describes entities and processes involved in producing and delivering or otherwise influencing that resource. Provenance provides a critical foundation for assessing authenticity, enabling trust, and allowing reproducibility.

Luc: Proposal on attributes - attributes are useful for interop with other provenance representations
... questions about constraints associated with attributes; propose to remove these from the data model
... we may revisit entity constraints in context of semantics

<Luc> Accepted: Constraints related to attributes will be dropped: (derivation-attributes, use-attributes, generation-affects-attributes)

Luc: second proposal: will not enter into detail of what attributes are "characterizing"
... will start on document rev with all accepted proposals

End of meeting.

- DRAFT -

Provenance Working Group Teleconference

03 Nov 2011

Attendees

Contents

Admin

F2F2 and F2F3

PROV-O

paq document

PROV-DM

Summary of Action Items

Scribe.perl diagnostic output