<trackbot> Date: 19 January 2012

<Luc> Scribe: satya

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

Luc: Discuss PROV-DM, identifiers, and if possible accounts

<Luc> PROPOSED: to accepted the minutes of Jan 12 teleconference

+1

Admin

<Paolo> +1

<Mike> +1

<tlebo> -1 http://www.w3.org/2011/prov/meeting/2011-01-12 404s

<smiles> 0 (absent)

<Luc> http://www.w3.org/2011/prov/meeting/2012-01-12

<dgarijo> +1

<stephenc> +1

<tlebo> +1

<Luc> Accepted: the minutes of Jan 12 teleconference

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

Luc: Reviewing outstanding actions
... 1. Write a blog on PROV-DM

<Luc> http://www.w3.org/2011/prov/wiki/Meetings:Telecon2012.01.05#prov-dm

<Luc> 185, 186, 187, 188, 189, 192, 193, 200

<tlebo> (the issues that are on Satya's action)

Luc: 2. Action for Satya to respond to outstanding issues

prov-dm

Luc: Please sign up for scribe duties

<Luc> http://lists.w3.org/Archives/Public/public-prov-wg/2012Jan/0073.html

<tlebo> missed it. Didn't realize there was a vote.

<smiles> yes, sorry, just been swamped, no other reason

<Paolo> mostly lost in the noise for me

Luc: Need to vote on issue at: http://lists.w3.org/Archives/Public/public-prov-wg/2012Jan/0073.html

<pgroth> +q

Luc: Will send a reminder to vote

<Luc> PROV-ISSUE-206: three proposals to vote on (deadline Jan 15th midnight GMT)

Paul: Indicate the content for voting in the subject line

Luc: Already listed in the subject line of the mail
... Next item - Paolo and me have been editing PROV-DM - specialization and alternateOf
... Reaching agreement on the transitivity of these constructs
... Have addressed all issues for the third release of DM

<Paolo> @stian are you available tomorrow -- new attempt to connect re: the collections setion

Luc: Outstanding issues have been listed at end of each section
... Propose to vote on release of DM as third working draft for next week

prov-o

Luc: Next item - PROV-O document

<tlebo> todo list that has been making progress is in meeting notes http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology_Meeting_2012-01-16

<dgarijo> +q

<dgarijo> @satya: that's great

Mike is joining the PROV-O team as co-editor

Mike: Working on blog post examples will soon post an update

<dgarijo> @satya : will we be discussing the best practices doc this monday too?

<Luc> Luc: Mike is joining the PROV-O team as co-editor

<dgarijo> +q

<pgroth> +q

Daniel: The PROV-O is becoming unwieldy with qualified involvement construct
... Would it make sense to have two resources?

<khalidbelhajjame> @Paul, I think we were thinking more about files, where people who are interested in a provo-light can use it

<tlebo> +q to say that we'll need "modules" for the prov-o "syntax" and prov-o "constraints" portions. So we'll need to tackele "modules" anywayy.

Paul: Recommend have both resources with same namespace instead of having two separate resources

<dgarijo> @tim: yeah, but the modules can share the namespace, right?

<dgarijo> I have to agree with paul on the namespace requirement.

<Zakim> tlebo, you wanted to say that we'll need "modules" for the prov-o "syntax" and prov-o "constraints" portions. So we'll need to tackele "modules" anywayy.

Luc: Agree with Paul for using single namespace for PROV-O
... May be premature to re-organize the ontology

<pgroth> +q

Tim: Is there a single approach for PROV-DM and PROV-O w.r.t. namespace

Luc: It is on the table for discussion

<Zakim> tlebo, you wanted to ask for clarity in namespaces among dm, prov-o, etc.

Luc: Proposal to have namespace for different PROV resources (XML, OWL etc.)

<tlebo> +1 tracking it as an issue.

Paul: This should be raised as an issue

<Luc> it's already a comment in prov-dm

<dgarijo> @paul: sounds good as a reminder.

<tlebo> I've pulled some notes on this last time I heard about it: http://www.w3.org/2011/prov/wiki/PROV-O_URI_namespace

Identifiers in Prov-dm

<Luc> http://www.w3.org/2011/prov/wiki/ProvenanceOfW3CReport

<Luc> http://www.w3.org/2011/prov/wiki/ProvenanceOfW3CReport#Discussion_about_Identifiers

Luc: Created an example to highlight problematic issues in DM w.r.t to identifiers

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

Luc: Agree on some of the issues related to identifiers - enumerated proposals

Simon: What has been proposed w.r.t common relations in DM, for example wasRevisionOf and usage events

<pgroth> +q

Simon: For example, A is revisionOf B then identify generation of A?

<pgroth> to respond to simon

Luc: Elicit feedback on example first

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

<dgarijo> satya: what happens if in account1 I make all the assertions in account2. What would be the relation between them?

<Luc> entity(w3:WD-prov-dm-20111215, [ prov:type="html4" ])

@Daniel: thanks Daniel!

<dgarijo> @satya: np

<Luc> http://dvcs.w3.org/hg/prov/raw-file/default/model/examples/w3cpub/w3c-publication2.prov-asn

<Luc> entity(w3:WD-prov-dm-20111215, [ prov:type="WD" ])

<Luc> http://dvcs.w3.org/hg/prov/raw-file/default/model/examples/w3cpub/w3c-publication1.prov-asn

<Luc> entity(w3:WD-prov-dm-20111215, [ prov:type="WD", prov:type="html4" ])

<pgroth> it's fine

<pgroth> nothing happens

<Paolo> @satya: no special action needed I guess

<khalidbelhajjame> +q

<dgarijo> @satya: well, accounts can be redundant, right?

<pgroth> +1 to MacTed

<Paolo> @satya that's still fine we don't draw any conclusion from comparing the content of two accounts

<pgroth> maybe we can get a summary

<pgroth> of the problem

MacTed: Misapprehension about the problem and the possible solution

<khalidbelhajjame> -q

<Luc> http://www.w3.org/2011/prov/wiki/ProvenanceOfW3CReport

<GK> (It seems to me that the different accounts per http://www.w3.org/2011/prov/wiki/ProvenanceOfW3CReport correspond to differing epistemological positions)

<Paolo> @Luc I suggest that we move to the specific proposals

@GK: I agree

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

<Luc> There is a requirement that *all* objects of discourse are identifiable and have an identifier.

Luc: Identifiers seem to denote entities and at other places it identifies records

<MacTed> "have an identifier" -- better "have one or more identifiers"

<Zakim> GK, you wanted to ask if we need *all* elements in domain of discourse to be identified. E.g. In RDF we have blank nodes.

GK: Not all elements in domain of discourse may be identifiable

Luc: Current DM states that all entities have to be identifiable but may not have an identifier?
... If there is no identifier, then how can they be referred to?

GK: Should identifiers in DM be same identifiers as in other serializations?

<GK> Also, I asked if "entity records" are *in* the domain of discourse

@GK: If they are blank node identifiers what is their state outside of the specific RDF document?

<GK> I agree that *accounts* are in the domain of discourse. Are these same as records?

<stainPhone> generation event of entity is akwsys identifiable as long as the entity is

Luc: Need to discuss the provenance of accounts, hence accounts are part of universe of discourse

<stainPhone> (have to go)

<MacTed> once I have an Account, I have an AccountCreator -- tho the latter may be "unknown" in all senses other than <entity> "CreatedAccount" <account>

<Luc> entity(w3:WD-prov-dm-20111215, [ prov:type="WD" ])

<khalidbelhajjame> +q

<GK> (Seems to me: an entity record (not in domain of discourse) introduces a name that denotes the described entity in domain of discourse. Unclear to me is whether this name is expected to carry though into any concrete representation (e.g. RDF)) I think this is what Luc is saying about proposal 1.

Simon: There are objects in domain of discourse that may not have identifiers?

@Simon Sorry didn't get your example

<MacTed> "*All* objects of discourse MUST be identifiable, and MAY have one or more identifiers."

<Paolo> +q

@Mac: +1

<pgroth> +1 to MacTed

<dgarijo> @satya: he said that if a wasRevisionOf b, then the activity that generated a (revisionActivity) could not be identified

<GK> (@simon: I think DM may require/introduce identifiers for things that are described by DM records, not necessarily for everything in domain of discourse)

<Paolo> +q to answer "yes" to Simon: necessary existence of an entity is not enough to know its identifier, clearly

@Daniel: thanks! :)

<GK> @macted +1

MacTed: If something is identifiable does not mean they have an assigned identifier

<dgarijo> @simon: so maybe in your example the activity is identifiable, but we may not know the identifier.

<dgarijo> @MacTed +1

Luc: If there is no identifier then how are they referred

<pgroth> you give it one

@Mac, Paul: exactly +1

<Paolo> @simon: may have not been clear earlier: we do not know the id of entities that must exist (existential quantifier) but are not the object of any assertion, i.e., they remain implicit

<smiles> @dgarijo OK, so under what you are suggesting, you mean that if we want to refer to the activity implied by a revisionOf relation, then we would introduce an identifier at that point?

<MacTed> "*All* objects of discourse MUST be identifiable, and MAY have one or more identifiers. For purposes of discourse, an identifier SHOULD be assigned to the object of discourse." ?

<GK> "There is a requirement that *all* objects of discourse are identifiable, and that entity records use or introduce an identifer for the objects described"

<pgroth> @MacTed good start

<dgarijo> @smiles: I guess so. You can invent one.

<tlebo> yea!

<pgroth> +1 to Graham

<tlebo> +1 @gk

@GK: +1

<GK> @macted - yes

<Paolo> @GK only entity records? how about activities etc.

<GK> @paolo ack.

<dgarijo> @Paolo: that a nice point: do we have an "alternateOf" for activities?

MacTed: Implied objects may not have identifiers

<Luc> There is a requirement that *all* objects of discourse are identifiable, and that entity records use a new identfier or introduce an existing identifer for the objects described

<GK> + "Implied entioties need not have an assigned idnetifier."

<tlebo> ?

<Zakim> MacTed, you wanted to say identifiers may be context-specific, temporary, "that thing there"...

<Zakim> Paolo, you wanted to answer "yes" to Simon: necessary existence of an entity is not enough to know its identifier, clearly

<Luc> There is a requirement that *all* objects of discourse are identifiable, and that entity records use a new identfier or introduce an existing identifer for the objects described

<tlebo> "use a new identfier or introduce an existing identifer" --> "introduce a new identifier or reuse an existing identifer"

<Paolo> @daniel: possibly, but this ID issue is not brought in "just" for alternateOf, rather it's a general principle that we decide to adopt

<tlebo> There is a requirement that *all* objects of discourse are identifiable, and that entity records introduce a new identifier or reuse an existing identifier for the objects described.

<Luc> There is a requirement that *all* objects of discourse are identifiable, and that entity records reuse an existing identifier or introduce a new identifier for the objects described

<tlebo> "provenace records"

<Luc> There is a requirement that *all* objects of discourse are identifiable, and that records reuse an existing identifier or introduce a new identifier for the objects described

<pgroth> +q

@Tim, +1

<MacTed> "*All* objects of discourse MUST be identifiable. Object descriptions MUST use an identifier (either reusing an existing identifier, or introducing a new identifier) for the objects described."

Paolo: Any object should be subject to the identifier rule

<MacTed> (maybe change the second MUST to SHOULD)

<MacTed> (yes, I'm deliberately removing extraneous "there is a requirement that" wording from the text.)

Paul: Concerned that implicit elements may have identifiers and proposal 1 does not cover that?

<smiles> Agreed with Luc - as long as record is explicit, it is fine

<tlebo> +1 to including RFC2119 (like MacTed's)

<Zakim> GK, you wanted to note we haven't said anything about the nature of these identifiers. Thisis OK, but we haven't licensed any further assumptions yet.

<Luc> There is a requirement that *all* objects of discourse are identifiable, and that records reuse an existing identifier or introduce a new identifier for the objects described

GK: Concerned that it does not lead to additional entailments from this proposal?

<smiles> +1

<Paolo> +1

<khalidbelhajjame> +1

<Luc> PROPOSAL: There is a requirement that *all* objects of discourse are identifiable, and that records reuse an existing identifier or introduce a new identifier for the objects described

<MacTed> I need a clear PROPOSAL to vote on...

+1

<GK> +1

<khalidbelhajjame> +1

<MacTed> -1

<smiles> +1

<tlebo> -1 should include RFC2119 terms

<MacTed> the language semms deliberately obscuring

<dgarijo> +1, although I liked more MacTed's

<sandro> +1

<sandro> (agreed -- this is not the final wording, just the idea.)

<tlebo> +1 for intent (please add RFC2119 like MacTed's)

<zednik> +1

<Paolo> so the vote is on the "general principle" only?

<GK> @macted - I think the intent is same as what you said: "*All* objects of discourse MUST be identifiable.  Object descriptions MUST use an identifier (either reusing an existing identifier, or introducing a new identifier) for the objects described."

<MacTed> PROPOSAL: "*All* objects of discourse MUST be identifiable.  Object descriptions MUST use an identifier (either reusing an existing identifier, or introducing a new identifier) for the objects described."

<GK> +1

I thought we were voting for the final version of the proposal?

<sandro> sandro: I don't think RFC 2119 language works unless it's clear who/what is constrained to follow it.

<tlebo> +1 to acknowledging these two have the intent and moving on.

Paolo: Is the current formulation reconciling record and entity?

<pgroth> whatever you want

Paolo: Which identifier needs to be re-used?

<pgroth> @Paolo: whatever you want

<tlebo> @paolo - choosing to reuse an identifeir is up to the asserter based on its properties.

<pgroth> @paolo whatever you want

Paolo: What criteria is used to choose to re-use an identifier?

<khalidbelhajjame> +q

<tlebo> @paolo - choosing to reuse an identifeir is up to the asserter based on its properties.

<MacTed> PROPOSAL: "*All* objects of discourse ("entities") MUST be identifiable by all participants in discourse.  Object descriptions ("entity records" and otherwise) SHOULD use an unambiguous identifier (either reusing an existing identifier, or introducing a new identifier) for the objects described."

<tlebo> alternateOf!

<GK> The point is, I think, when the intent of assertion is to refer to something already described, to re-use the identifier already used. This is how names work, no?

<tlebo> (not complementOf)

<tlebo> @GK, "GK, who?" ;-)

Khalid: Should use same identifier for same entity?

<Paolo> @tlebo makes sense, in principle

<GK> @tlebo the same GK as referred to previously :)

<MacTed> if you know someone used a name for the thing (and what that name is), then sure, you might choose to reuse their naming. what if you don't know they did, nor what name they chose

<MacTed> ?

Khalid: Not including a criteria for re-using identifier is fine

<Luc> PROPOSAL: "*All* objects of discourse ("entities") MUST be identifiable by all participants in discourse. Object descriptions (e.g., "entity records" and otherwise) SHOULD use an unambiguous identifier (either reusing an existing identifier, or introducing a new identifier) for the objects described."

<dgarijo> +1

<GK> +1

<MacTed> +1

<Paolo> +1

<smiles> +1 (for intent)

<jcheney> +1

<khalidbelhajjame> +1

<tlebo> +1 to intent, just like Luc's

<dgarijo> satya: is object description just an example?

+1

<zednik> +1

<sandro> +1

<Luc> ACCEPTED: "*All* objects of discourse ("entities") MUST be identifiable by all participants in discourse. Object descriptions ("entity records" and otherwise) SHOULD use an unambiguous identifier (either reusing an existing identifier, or introducing a new identifier) for the objects described." (intent)

<Luc> PROPOSAL: Generation and Usage events also belong to the universe of discourse. So they should be given identifiers (as per proposal 1)

<jcheney> This is in line with the current formal semantics.

<smiles> I agree, it seems a natural consequence of proposal 1

<pgroth> @smiles +1

<dgarijo> the only problem I see is the identifiers that would be given in 2 accounts that are describing the same process with different identifiers.

<GK> I think there are two remaining areas for discussion: (a) what are the objects in the domain of discourse, and (b) what identifiers may be used in DM and how to the relate to, e.g., URIs used in RDF.

<pgroth> @GK - Nice point

<smiles> @dgarijo Isn't that just an unavoidable problem for anything identifiable?

<tlebo> @dgarijo, what is wrong with "identifiers that would be given in 2 accounts that are describing the same process with different identifiers." ?

<MacTed> I think that "objects in the domain of discourse" actually include Events (Generation, Usage, and otherwise), Derivations, Notes, and otherwise

<dgarijo> how would I say that both activities are the same? owl:sameAs?

<tlebo> prov:alternativeOf or owl:sameAs

<GK> @MacTed - I think so too.

<dgarijo> alternativeof is for entities ;)

<MacTed> 2 accounts using 2 identifiers for the same entity is an implementation issue -- i.e., someone needs a Reasoning Engine

<MacTed> to handle the owl:sameAs (or other) relationships

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/"entity records" and otherwise/e.g., "entity records" and otherwise/
Found Scribe: satya
Inferring ScribeNick: satya
Default Present: pgroth, Paolo, Luc, Satya_Sahoo, AlexHall, tlebo, dgarijo, sandro, jcheney, MacTed, +1.518.633.aaaa, khalidbelhajjame, +44.789.470.aabb, GK
Present: pgroth Paolo Luc Satya_Sahoo AlexHall tlebo dgarijo sandro jcheney MacTed +1.518.633.aaaa khalidbelhajjame +44.789.470.aabb GK
Agenda: http://www.w3.org/2011/prov/wiki/Meetings:Telecon2012.01.19
Found Date: 19 Jan 2012
Guessing minutes URL: http://www.w3.org/2012/01/19-prov-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.