Provenance Working Group Teleconference

26 Sep 2011

See also: IRC log


Satya_Sahoo, +1.315.330.aaaa, stain, +44.238.059.aabb


<trackbot> Date: 26 September 2011

<tlebo> wiki for this meeting: http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology_Meeting_2011-09-26


<tlebo> http://dvcs.w3.org/hg/prov/raw-file/2b514f651c03/ontology/examples/ontology-extensions/crime-file/diagrams/2011-09-26-tlebo/crime.owl.png

<Luc> what is provenanceContainer in entity? I don't understand the notation used.

file: ///home/stain/stuff/src/provenance-wg/prov/ontology/ProvenanceFormalModel.html#provenancecontainer I guess

that is a subclass

<tlebo> TODO: break all visual abbreviations - show rdfs:subClassOf directly.

Luc does not like that a ProvenanceContainer is a subclass of Entity

<tlebo> TODO: ensure that the ProvenanceContainer is no longer an Entity. (Tim)

<Luc> it's not that Luc doesn't like. It does not match the data model.

In PROV-DM, an entity expression is a representation of an identifiable characterized thing.

so is the provcontainer not one of those?

<Luc> no

<tlebo> luc: top level classes: ProvenanceContainer, Account, Entity, ProcessExecution


<Luc> http://dvcs.w3.org/hg/prov/raw-file/default/model/ProvenanceModel.html#ProvenanceContainer

<satya> This is PROV-ISSUE-66

<tlebo> paolo: we are not "directly" asserting provenance of ProvenanceContainer

<khalidbelhajjame> I think we should just stick to the data model as much as possible

<dgarijo> @khalid: yes, of course. But if the model is designed for asserting provenance, then it should provide the mechanisms to assert provenance about provenance records too!

<khalidbelhajjame> +q

<Luc> satya, there is a type error!

<tlebo> (the objective of the diagram was to illustrate how a third party could extend PROV - does anyone have feedback on how well/poorly this is conveyed?)

<dgarijo> could you provide the link to the diagram, please? I joined late :(

tlebo: I think it's good, but I wonder if we could show extension by subproperties or subclass

<tlebo> http://dvcs.w3.org/hg/prov/raw-file/2b514f651c03/ontology/examples/ontology-extensions/crime-file/diagrams/2011-09-26-tlebo/crime.owl.png

just adding new attributes to an prov:Entity is merely how entity is to be used anyway..?

<tlebo> @stian: news:Entity subclass prov:Entity , THEN assert the domains on news:Entity ?

<Luc> you can take multiple perspective on a provenance container: it's a curated set of assertions, it's a set of triples in a triple store, it's a 'platonic' set of triples, it's something in a database.

so Journalist subclass of Agent, or Publishing subclass of ProcessExecution might be interesting. But then you are very close to the danger of touching on roles..

<Luc> Each of them can be modelled by an entity expression.

<Luc> An apple is represented by an entity expression, which involves selecting some attributes

<khalidbelhajjame> -q

http://dvcs.w3.org/hg/prov/raw-file/default/model/ProvenanceModel.html#expression-Account says

Note: Currently, the non-terminal asserter is defined as URI. We may want the asserter to be an agent instead, and therefore use PROV-DM to express the provenance of PROV-DM assertions. The editors seek inputs on how to resolve this issue.

is this what we've briefly touched on?

<Luc> yes stain

<dgarijo> I have to agree with Satya

<dgarijo> I don't see the difference either.

<dgarijo> if a prov container can be set of statements, and we are able to refer to them and assert triples using them as source, then it has a identifier.

<Paolo_> the corollary of what Luc is saying, i.e., viewing a provenance container as a "thing in the world", is that you don't need to make ProvenanceContainer a subclass of Entity in order to do "meta-provenance"

so Luc's point is that there (could be) other attributes you need to specify, like the location the provenance container is stored in, or description of its content, etc - many entities

<dgarijo> +q

<khalidbelhajjame> En entity is a characterization of a thing, provenance container is not necessarily a characterization of a thing.

mm.. so there's nothing saying a ProvContainer *is not* an entity, but it would probably not be a good entity on its own because there are many ways to describe such meta-provenance

<khalidbelhajjame> A provContainer can be a characterization, but not necessarily one.

<dgarijo> -q

<dgarijo> ok, now it is clear for me. So basically Prov container and Entity are not disjoint, but not necessarily the former subclass of the latter.

and t might be misleading to say it is always an entity, because it hints that you *should* do meta-provenance in only a single way

<dgarijo> :)

<tlebo> provenanceContainer: is a list of accounts, namespace definitions, and a collection of provenance assertions.

so provenanceContainer() in the model is rdf:RDF (or equivalent headers) in RDF

<tlebo> JSON needs ProvenanceContainer, while RDF/OWL does not need it "because it comes for free".

<dgarijo> @tim: but it depends on the way you implement it: if you select named graphs, then your provenance container could be the named graphs.

mm.. and then you can do RDF->model->JSON->model->RDF and somewhat at least don't mess up namespaces and stuff

<tlebo> TODO: class diagram (Khalid)

<Luc> how is this group proposing to model accounts?

<khalidbelhajjame> http://dvcs.w3.org/hg/prov/raw-file/default/ontology/ProvenanceFormalModel.html

<satya> http://dvcs.w3.org/hg/prov/raw-file/default/ontology/ProvenanceFormalModel.html#specialization-of-provenance-ontology-classes

<dgarijo> @Luc: should they be a subclass of provenance container?

@Luc I think we are still thinking of that as named graph / separate OWL/RDF resources per account - due to how identifiers depends on the account in the abstract model

<dgarijo> since they are a collection of provenance assertions too.

<Luc> yes, probably name graph would be the way ... but then this does not show in OWL?

<tlebo> stian: show the other subclasses of the extension.

<tlebo> (TODO: Tim show this)

<khalidbelhajjame> @Luc, yes I agree, we should be as explicit as possible

<dgarijo> and what about having account as subclass of prov container? I think it could be fine.

<Luc> @daniel, an account would then inherit prov container's properties ... but accounts have no namespace, and no index of accounts

<Paolo_> @Daniel yes that would be reasonable -- something we discussed in the model

<tlebo> the news crime file example: http://dvcs.w3.org/hg/prov/raw-file/2b514f651c03/ontology/examples/ontology-extensions/crime-file/instances/example-1/crime.ttl.vertical.png

I would suggest to do role example with a process execution with 2 or more uses or 2 or more generations

so that you see why you would want a role

or 2 or more agents

I have some in http://dvcs.w3.org/hg/prov/raw-file/default/ontology/examples/ontology-extensions/workflow/prov2.ttl

<Luc> i will have to go, sorry!

<dgarijo> bye!

<satya> ok bye Daniel

<satya> http://dvcs.w3.org/hg/prov/raw-file/default/ontology/ProvenanceFormalModel.html#entity

those notes don't match the exxmple

<dgarijo> no, I wasn't leaving, just saying bye to Luc :)

<satya> oh ok :)

<satya> bye Luc

<tlebo> +1

<tlebo> satya: we let the conceptual model become the logical model.

<jcheney> Got to go.

<dgarijo> see you, jcheney.

<Paolo_> guys I really have to go

<dgarijo> see you, paolo!

<Paolo_> bye

<tlebo> trackbot, end telcon

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.136 (CVS log)
$Date: 2011/09/26 17:18:54 $

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)

No ScribeNick specified.  Guessing ScribeNick: stain
Inferring Scribes: stain

WARNING: No "Topic:" lines found.

Default Present: Satya_Sahoo, +1.315.330.aaaa, stain, +44.238.059.aabb
Present: Satya_Sahoo +1.315.330.aaaa stain +44.238.059.aabb
Found Date: 26 Sep 2011
Guessing minutes URL: http://www.w3.org/2011/09/26-prov-minutes.html
People with action items: 

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

WARNING: No "Topic: ..." lines found!  
Resulting HTML may have an empty (invalid) <ol>...</ol>.

Explanation: "Topic: ..." lines are used to indicate the start of 
new discussion topics or agenda items, such as:
<dbooth> Topic: Review of Amy's report

[End of scribe.perl diagnostic output]