Provenance Working Group Teleconference -- 13 Sep 2012

<trackbot> Date: 13 September 2012

<pgroth> trackbot, start telcon

<trackbot> Meeting: Provenance Working Group Teleconference

<trackbot> Date: 13 September 2012

<pgroth> Scribe: Satya Sahoo

Admin

<pgroth> http://www.w3.org/2011/prov/meeting/2012-09-06

<satya> Paul: Agenda items: updates on F2F, PROV-O admin issues, outstanding issues, and implementation reports

<pgroth> Minutes of the September 6, 2012 Telecon

<satya> +1

<ivan> +1

<pgroth> approved: Minutes of the September 6, 2012 Telecon

<satya> Paul: Open action on Paulo, Paul still open

<satya> Paul: Action item regarding XML schema definition

<satya> Hook: Completed the XSD

<satya> Paul: Action item on Hook closed

F2F4

<satya> Paul: Please sign up for scribe

<pgroth> http://www.w3.org/2011/prov/wiki/F2F4

<satya> Paul: Next agenda item regarding 4th F2F

<satya> Paul: Please sign for the F2F either remotely or in person

<satya> Paul: Question regarding F2F?

Prov-o telcon

<satya> Paul: PROV-O telcon was important for development of the ontology, currently responding to the comments

<satya> Paul: Discussed with prov-o team to roll in the prov-o discussion items into the regular PROV call

<Zakim> Luc, you wanted to suggest prov-xml team could use the prov-o slot

<satya> Luc: PROV-XML team may want to use the PROV-O time slot on Monday for discussion

<satya> Stephan: Agree if rest of the PROV-XML team agrees

<hook> sounds good

<satya> Paul: PROV-XML will use the monday timeslot for meetings

PROV-Constraints Dissemination

<pgroth> http://www.w3.org/TR/prov-constraints/

<satya> Paul: Next agenda item: PROV constraints dissemination

<satya> Paul: Congrats to the editors of PROV constraints!

<Luc> http://www.w3.org/blog/SW/2012/09/12/last-call-constraints-of-the-provenance-data-model-2/

<satya> Paul: Paolo, Luc created a blog post describing the PROV constraints

<satya> Paul: Please share the blog with others

<satya> Paul: Can someone volunteer to share this with various maling lists

<satya> maling/mailing

<pgroth> ivan to semweb list

<pgroth> satya to hclrs

<satya> Paul: Luc can share with provenance challenge

<pgroth> luc provenance challenge mailing list

<satya> Paul: James to share with Dagstuhl mailing list

<pgroth> james cheney to mailing list

Resolving Issues

<satya> Paul: Issues raised for PROV-DM (primarily from Robert@Mayo clinic)

<satya> Luc: Reviewed the feedback for comments that require changes to the data model

<satya> Luc: Drafted responses to these specific issues

<satya> Luc: ISSUE 532 regarding roles, to allow roles for more relations

<Luc> http://lists.w3.org/Archives/Public/public-prov-wg/2012Sep/0141.html

<satya> Luc: WG decided against this as there was no satisfactory resolution

<satya> Luc: Protocol to follow to resolve these issues - vote or put it up for discussion on a wiki page

<ivan> yes

<satya> Paul: If there are no objections raised by a particular date, then confirm with the person who raised the issue

<satya> Ivan: It is good to have a confirmation to the proposed solution, but it may not be always possible

<satya> Paul: Let the WG consider the response by Monday and then get back to the person raising the issue

<satya> Luc: Where should the response be posted?

<satya> Ivan: In the tracker

<satya> Ivan: The response should include links to the person's confirmation

<satya> Luc: ISSUE 525: Specialization and Alternate to be subtype of Influence

<satya> Luc: WG resolution on this issue

<pgroth> note, working group members should respond to these suggestions by monday

<satya> Luc: Questions/comments on ISSUE 525

<satya> Luc: Include attributes for these relations, create patterns similar to the relations

<satya> Luc: There was decision against this approach

<satya> Paul: Should include this discussion in the response

<satya> Paul: Add a section to the wiki page for the issue to include updates

<pgroth> ivan?

<satya> Luc: ISSUE 507: wasAssociatedWith in DM, PROV-O includes this property and also an inverse relation - to include the inverse relation in DM

<satya> Luc: In relational design, we do not include inverse relations. PROV-O includes inverse relations due to OWL/RDF

<satya> Ivan: Understands Luc's point, Is it necessary to have an inverse relation?

<satya> Ivan: It is usual to have this pattern in an ontology, but it may lead to complexity

<satya> Mac: The notion of what is a relation and what is its inverse is arbitrary

<satya> Mac: So, both should be there

<satya> Stephen: There was a lot of discussion in PROV-O regarding this, it is currently a suggestion for the inverse properties (to ensure consistent naming)

<satya> Stephen: The inverse is not always explicitly specified

<satya> Ivan: Not sure what does it mean - are they part of the OWL file?

<satya> Tim: They are not, the inverse are in a separate OWL file

<satya> Ivan: Not convinced, not having them at all is a better solution

<satya> Tim: PROV-O tried to reach a compromise and the current solution is after long discussion

<satya> Luc: PROV-O does not include inverse relation, hence it is normative - so PROV-O and DM are aligned

<satya> Luc: Will rephrase the answer

<Luc> http://www.w3.org/2011/prov/track/issues/504

<ivan> issue-504?

<trackbot> ISSUE-504 -- Data Model Section 2.2.2 -- open

<trackbot> http://www.w3.org/2011/prov/track/issues/504

<satya> Luc: ISSUE-504, regarding bundles and collections

<satya> Luc: WG has discussed this issue on multiple points - WG resolved not to have identifiers on provenance assertions, bundles allow assertion of identifiers and make statements about bundles

<satya> Luc: WG considers this as a good compromise

<satya> Paul: Include a change in definition of collections?

<Luc> http://www.w3.org/2011/prov/track/issues/503

<satya> Luc: ISSUE 503: regarding agents that adopt a plan

<satya> Luc: wasAdoptedBy should be a new definition in PROV to allow plan to be assigned to agent without use of activity

<satya> Luc: Responded to Paul separately, not sure what is to be done here

<dgarijo> +q

<satya> Daniel: Does he have a use case that would justify the introduction of this new relation?

<satya> Luc: Not to my knowledge

<satya> Daniel: We have enough relations to cover this use, so would not vote to have a new relation

<dgarijo> that makes sense

<satya> Paul: Is this relation an easy extension to existing PROV model?

<satya> Tim: Would it be a property chain or qualified association based rule?

<satya> Paul: This suggestion can be included in the response

<satya> Luc: Will edit the wiki page tomorrow and send mail - give time till Tuesday

<satya> Paul: Any objections to have responses to this discussion by Tuesday?

<pgroth> accepted: to have responses to the issues in the agenda by tuesday

Subactivity

<satya> Paul: Next agenda item is discussion about subactivity

<pgroth> http://www.w3.org/2011/prov/track/issues/447

<satya> Paul: ISSUE 447, several request to introduce a sub activity relation in the model followed by mailing list discussion

<satya> Paul: It may be a good idea according to some members, but others have reservations that introduction of the new relation will have repurcussions across model and constrains (some of which may have not be considered at all)

<satya> Paul: One proposed option is use dc partOf relation

<satya> constrains/constraints

<dgarijo> I think we should go for dcterms:hasPart.

<satya> Paul: Whether to include or leave out this?

<satya> @Daniel +1

<Paolo> I expressed my view on the list -- I'd rather leave it out of the model

<satya> Stephan: Can this be included in the best practices document? But not include this in specs

<satya> Luc: Daniel please clarify proposed use of dcterms:hasPart

<pgroth> +q

<satya> Daniel: Keen to have it in a document, but not in ontology - is there a proposal for a best practices document as it would allow inclusion of some issues currently pending in PROV-O also

<dgarijo> http://dvcs.w3.org/hg/prov/raw-file/tip/bestpractices/BestPractices.html

<Luc> @zednik: should it be an FAQ?

<satya> Stephan: (answering to Luc) Last F2F discussed creation of best practices document - it will be a note and not a spec

<satya> Stephan: It will be a kind of FAQ, maybe part of FAQ

<MacTed> +1 "this is a common need; here's how best to satisfy it with what we've done"

<Luc> proposed: subactivity relation will not be included in normative documents, but instead would be described in light weight document, such as FAQ/best practice

<satya> Paul: We have identified the DC document as a best practice document, strongly support a FAQ document

<satya> Paul: FAQ will sidestep the complications associated with a spec, so this FAQ can also be updated

<Luc> proposed: subactivity relation will not be included in normative documents, but instead would be described in light weight document, such as FAQ/best practice

<dgarijo> +1

<zednik> +1

<ivan> +1

<MacTed> +1

<satya> +1

<SamCoppens> +1

<TomDN> +1

<Paolo> +1

<hook> +1

<satya> Tim: +1

<jun> +1

<Dong> +1

<Luc> accepted: subactivity relation will not be included in normative documents, but instead would be described in light weight document, such as FAQ/best practice

<satya> Paul: Will setup FAQ that can also be referenced when responding to feedback

<pgroth> ACTION: setup faq [recorded in http://www.w3.org/2012/09/13-prov-minutes.html#action01]

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

<pgroth> ACTION: pgroth setup faq [recorded in http://www.w3.org/2012/09/13-prov-minutes.html#action02]

<trackbot> Created ACTION-115 - Setup faq [on Paul Groth - due 2012-09-20].

<ivan> http://www.w3.org/2001/sw/wiki/Main_Page

<satya> Ivan: There is a SW activity wiki, which can be used after WG formally closes - create a separate section on the wiki that can be continously used

<satya> Paul: Nice idea

Implementation Report

<satya> Paul: Will set up a PROV subsection on SW wiki

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

<satya> Paul: Thanks to Dong and Stephan to creating the implementation document

<satya> Paul: The skeleton of the document mirrors Stephan's questionnaire

<satya> Paul: Reviewing the implementation document

<satya> Paul: Comments/questions about the structure of the implementation document?

<ivan> (looked fine to me...)

<satya> Tim: Are there instructions to allow responders to add to this document?

<satya> Paul: Part of the questionnaire

<satya> Stephan: Questionnaire is a spreadsheet so should be easy to map to document

<satya> Paul: Have a question about constraint section, it currently does not have a test case for implementation of the constraints

<satya> Paul: Should there be a test harness for this - concern that this will require significant work

<satya> Paul: Question is whether it is needed and who will do it?

<satya> Ivan: Not sure about test case environment

<satya> Paul: This environment allows testing a set of provenance assertions to be valid or invalid

<satya> Ivan: Nice to have an environment to make testing easier, but making a complete test environment requires lot of effort so not sure it can be done

<satya> Luc: I am implementing a validator and it involves lot of work - tests unification algorithm and requires lot of test cases (excludes temporal constraints)

<satya> Luc: Does not guarantee complete coverage and not sure when a stable state test cases can be shared with rest of WG

<satya> Luc: So, not sure how the test cases can be re-used by others

<satya> Luc: May require development of a common API, so not an easy task

<Dong> @zednik PROV-CONSTRAINTS now has 58 in total, I think the questionnaire still uses the constraints from the previous version of the document.

<satya> @Paul: sorry, I have to leave now can someone take over as scribe

<pgroth> ok

<zednik> @dong, thanks I'll look into that

<pgroth> can someone scribe?

<dgarijo> I can

<pgroth> thanks

<pgroth> scribe: dgarijo

<satya> thanks Dani!

Ivan: we need test cases when we can infer additional things. If it can be automatized, good.

Luc: the complete normal form is not necessarily represented in Prov.

Ivan: that is a lot of work. This formats may have their own live after the group finishes

Paul: even if we divide ourselves, we can't guarantee complete cover

<stain> hm, how 10 minuets becomes an hour.. I missed the whole thing!

Ivan: that's different. Test cases are never 100% complete (corner cases not covered). It should be possible to come up with a collection of test cases that the group supports. If not, we have a problem

<stain> is it still going on? Where in the agenda are we?

Ivan: everybody will have to have his own test cases and be able to report them back to us

@stain: we are discussing the implementation

<stephenc> SPARQL working group have a nice vocabulary for defining their test cases.

pgroth: the next thing to do is to find how to define the test cases.

Ivan: there are many other groups that have been doing this. It may be worth taking a look

<Dong> ok, sure

pgroth: stephen, dong would you take a look at that?

stephen: ok

pgroth: we are out of time
... we'll catch up online

<Dong> thanks, bye

<pgroth> trackbot, end telcon

- DRAFT -

Provenance Working Group Teleconference

13 Sep 2012

Attendees

Contents