Provenance Working Group Teleconference -- 11 Aug 2011

<trackbot> Date: 11 August 2011

<pgroth> Scribe: Ryan Golden

<stain> not many today!

<stain> Zakim: ??P34 is me

<Vinh> +1.937.343.aaff is me

<pgroth> http://www.w3.org/2011/prov/meeting/2011-08-04

<pgroth> PROPOSED to accept the minutes of 04 Aug telecon

<satya> +!

<Curt> +1

<jcheney> +1

<satya> +1

<kai> +1

<khalidbelhajjame> +1

<Yogesh> +1

<stain> +1 (not there, but good)

<smiles> +1

<Edoardo> +1

<pgroth> Scribe: Helena Deus

<StephenCresswell> +1

<olaf> +1

<GK> +1

<Lena> minutes approve

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

<pgroth> http://www.w3.org/2011/prov/wiki/W3C_Implementation_Stakeholder_Questionnaire_2nd_Iteration

<pgroth> Helena: discussing stakeholder questionnaire

<Helena> http://www.w3.org/2011/prov/wiki/W3C_Implementation_Stakeholder_Questionnaire_2nd_Iteration

<stain> Helena: Came across several concerns, need guidelines. Representation of provenance.

<pgroth> thanks stain

<stain> Helena: First basic description, second level more on how data was produced, but not everything, third level is complete provenance

<stain> Helena: Use this to create unit test, not yet formulated.

<stain> Helena: Need to contact existing stakeholders if it is OK to publish data

<pgroth> +q

<satya> +q

<MacTed> throwing a minor tweak (pre-formed answers) into the "may we share your free-text?" boilerplate...

<pgroth> +q

<stain> Satya: Last bit of form asks 'ow do you produce provenance', should we also ask 'how do you collect provenance' ?

<Lena> satya: should we also keep track of the tools are used to track provenance

<MacTed> http://www.w3.org/2011/prov/wiki/W3C_Implementation_Stakeholder_Questionnaire_2nd_Iteration#Email_to_contact_stakeholders_regarding_privacy_of_questionnaire_responses

<Lena> satya: had model telco on monday

<Lena> satya: adding query of example scenario

<Lena> satya: adding diagrams to illustrate constructs of the ontology (help users to understand axioms in the ontology)

<satya> http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology

<Lena> satya: calling for comments on "roles"

<Lena> satya: how should role be modelled in different scenarios?

<pgroth> +q

<jcheney> +q

<satya> James is referring to: http://dvcs.w3.org/hg/prov/raw-file/93fc4f6af76d/ontology/ProvenanceFormalModel.html

<Helena> jcheney: went over owl document and identified properties that could be formalized in the ontology

<satya> good point James! - I missed that point

<Helena> jcheney: how to represent n-ary relationships in owl

<Helena> pgroth: modify the provenance formal model using anchors

<GK> Re n-ary relations: http://www.w3.org/TR/swbp-n-aryRelations/

<pgroth> ACTION: jcheney put anchors into the Provenance Model [recorded in http://www.w3.org/2011/08/11-prov-minutes.html#action01]

<trackbot> Created ACTION-34 - Put anchors into the Provenance Model [on James Cheney - due 2011-08-18].

<stain> (but it does have anchors..? http://dvcs.w3.org/hg/prov/raw-file/93fc4f6af76d/ontology/ProvenanceFormalModel.html#time for instance)

<DGarijo> @GK wi didn't wat to declare the relationships as classes in the ontology

<DGarijo> @GK that's why we have used another approach by Satya

<pgroth> @james this should be identifying the html anchors in the provenance model to be used

<DGarijo> @GK I meant We didn't want, no Wi didn't wat...

<Helena> simon: location - what is it supposed to be? how should it be represented? formal model vs conceptual model inconsistencies

<khalidbelhajjame> q

<khalidbelhajjame> +q

<IlkayAltintas> How does the sensor working group identify location?

<Helena> simon: location in the pil model can be defined by another standards?

<stain> http://dvcs.w3.org/hg/prov/raw-file/default/model/ProvenanceModel.html#concept-Use

<GK> Thatwas *two* changes, wasn't it? Simple case "uses" -> "used". "Is used by" is also inverse relation.

<Helena> DGarijo: should have "uses"relationship instead of "usedBy" -> from Luc's comment on the wiki

<stain> http://dvcs.w3.org/hg/prov/raw-file/93fc4f6af76d/ontology/ProvenanceFormalModel.html#isusedby is in the past

<satya> @Ilkay: I think the sensor incubator group? uses OGC term for location

<Helena> pgroth: keep track of deviations between conceptual model and ontology

<Helena> khalidbelhajjame: many relationships in the conceptual model may need to be ignored

<Helena> khalidbelhajjame: that is why they appear to be inconsistent

<satya> @Khalid: I agree

<Helena> pgroth: provenance access and query document

<Helena> GK: drafted 2 proposals ofr an http inteface for prov discovery

<Helena> GK: first was not the interface developers were hoping to see

<Helena> GK: second proposal goes back to first principles and treats it as a REST interface

<Helena> GK: adding "concepts" section/review and refine terminology (converging on the model document term.)

<Helena> GK: added discussion about the issues of discovering provenance

<Helena> GK: link header -> recommend use of the anchor parameter

<Helena> pgroth: first version did not allow the define the "entity"

<Helena> pgroth: added the notion of target as well as a resource within the set of concepts

<Helena> pgroth: that is the anchor or url that can be used to find the provenance of something

<GK> http://dvcs.w3.org/hg/prov/raw-file/tip/paq/provenance-access.html

<Helena> Yogesh: multiple provenance URI and target URI -> will provenance uri have provenance about the target uri?

<Helena> pgroth: they could or could not have that infor

<Helena> pgroth: provenance information may or may not have provenance about a target uri

<Helena> Yogesh: does it make sense to define which provenance uri describe which target uri?

<Helena> pgroth: how to bind a target to a particular set of proveance information?

<Helena> pgroth: particulalry because a target could show up in many provenance info URI

<MacTed> bad psuedotriple -- { <provenanceURI> :primaryTopic <targetURI> }

<Helena> GK: html link element is hard to use because does not work like anchor

<stain> @pgroth, I think that makes sense, it's up to the provenance itself to say something about the target's provenance

<khalidbelhajjame> +q

<MacTed> or ... { <provenanceURI> :describes <targetURI> }

<MacTed> or ... { <targetURI> :describedBy <provenanceURI> }

<stain> vu

<Helena> khalidbelhajjame: provenance doc show say how to access the provenance of a particular entity -> would also allow connecting resource with entity

<Helena> pgroth: addressing the common case for access

<Helena> pgroth: in html -> what's the provenance of the page?

<Helena> pgroth: need to make that compatible with the notions from the model

<stain> but is <provenanceURI> allowed to talk about <provenanceURI> even if it does not say how it was made? Like <provenanceURI> pil:usedBy :someProcess

<Helena> pgroth: not clear in that case that a resource is an entity

<GK> (Expectation that the provenance itself would be explicit about what it applies to.)

<Helena> pgroth: html page could be a entity (target uri)

<GK> I already noted I think we need to add examples to make this area clearer.

<satya> sorry I did not understand - What is the difference between resource and entity?

<Helena> Yogesh: notion of resource in the access document -> should be raised as an issue with the conceptual model; the vocabulary was not sufficient to describe how to access provenance

<Helena> pgroth: that's not the role of the model

<GK> I think khalid has a point... I think we need to clarify what we're trying to say and revisit.

<satya> I agree with your point Khalid - but I am trying to understand why are we differentiating between resource and entity?

<Helena> Yogesh: had to use new notion "resource" to describe how to access provenance of things

<stain> I believe "resource" in PAQ doc is just a normal web resource, right?

<Helena> pgroth: need to connect to web architecture; use web architecture terms

<Helena> Yogesh: why use of target in http header (instead of anchor)?

<satya> Following on Khalid's point, can't we model web resource as type of entity?

<MacTed> +1 satya (that's what my queue+ is about...)

<Reza_BFar> Another option would be to create an indirection layer between a web resource and an entity and use something like XSL/XSLT to specify possible transformations as an extension... though this adds complexity which would be undesirable.

<Helena> call a context instead of a target?

<Helena> pgroth: reusing terminology from http specification, need ot make sure they "mean" what they "say"

<Reza_BFar> +1 on terminology from Paul.

<GK> HTTP uses "anchor" IIRC

<Helena> DGarijo: http uses target in the wrong way

<Helena> MacTed: a web resource is a resource; a subclass of an entity

<khalidbelhajjame> +1 for Jim

<Helena> MacTed: conceptual framework MUST maintain that!

<GK> How is resource a subclass of entity?

<Helena> MacTed: if we dont' treat it as so, we break everything

<GK> A rsource is a thing that can be named

<DGarijo> @Helena I wasn't the one talking :)

<Helena> MacTed: an entity may be of any type of class (e.g. web reousrce is an entity IF it can be named)

<GK> All of those entoties are also resources.

<Helena> MacTed: provenance is not just about where a web page comes from

<GK> I think we have crossed wires here. Not sure where right now.

<Helena> MacTed: e.g. a picture on a usb stick is not a web resource, but it does have provenance

<Helena> pgroth: provenance model document describes "entity"

<GK> I agree with most of what @MacTed said, except as noted...

<Helena> pgroth: in terms of the provenance model document

<Curt> back to BOB vs. entity?

<GK> I agree with @MacTed -- it's not the *web* resource that's special or distinct here...

<GK> (per se)

<Helena> MacTed: provenance of entities is intertwined

<Helena> MacTed: provenance should NOT be just about web resources

<Helena> MacTed: provenacne document should allow that

<jcheney> I thought we all agreed previously that "provenance should NOT be just about web resources"

<jcheney> (or at least it was generally the consensus)

<GK> @jcheney yes, I agree - and I think this discussion has become crossed if that's how it sounds

<satya> Lena: Why should entity be different in query and access document from provenance model?

<khalidbelhajjame> @paul, the provenoce model does not use resource, PAQ does

<GK> Sorry hunting mute button

<satya> @GK :)

<Reza_BFar> I agree with Paul on that the treatment in Query is different than the Model. The overwhelming use-case, IMHO, of how you "Query" a provenance provider will be through some http related interface hence lending itself to URI

<Zakim> GK, you wanted to suggest that paul and I should go and clarify between ourselves, then come back with revised proposal.document

<pgroth> http://dvcs.w3.org/hg/prov/raw-file/default/model/ProvenanceModel.html#shortcuts-and-extensions

<Helena> pgroth: propose shortcuts and extensions added to the provenance model document

<GK> ACTION: paul and graham to clarify what we mean by entity vs target vs resource [recorded in http://www.w3.org/2011/08/11-prov-minutes.html#action02]

<trackbot> Created ACTION-35 - And graham to clarify what we mean by entity vs target vs resource [on Paul Groth - due 2011-08-18].

<stain> which points are those..?

<satya> http://www.w3.org/2011/prov/track/issues/71

<Helena> pgroth: minimal list of what is useful to have (e.g. something that looks like attribution)

<satya> will add comments to mailing list thread

<pgroth> trackbot, end telecon

- DRAFT -

Provenance Working Group Teleconference

11 Aug 2011

Attendees

Contents

Summary of Action Items

Scribe.perl diagnostic output