Provenance Working Group Teleconference -- 13 Dec 2012

<trackbot> Date: 13 December 2012

<pgroth> trackbot, start telcon

<trackbot> Meeting: Provenance Working Group Teleconference

<trackbot> Date: 13 December 2012

<pgroth> Scribe: Paolo Missier

Admin

<pgroth> http://www.w3.org/2011/prov/meeting/2012-12-06

<pgroth> Minutes of Dec. 06, 2012

<tlebo> +1

<ivan> +!

<dgarijo> I wasn't there, +0

<TomDN> +1

<ivan> +1

0 (not present)

<smiles> +1

<hook> 0 (not present)

<jcheney> 0 (not present; I seem to be listed as both present & absent)

<Curt> 0 (not present)

<SamCoppens> +1

<pgroth> accepted: Minutes of Dec. 06, 2012 Telcon

pgroth: tlebo still working on his action
... we can close all issues around questionnaire
... stephan not on the call, we are closing the issues, we assume the questionnaires are done
... action 151 done. will elaborate. action-153 also done
... still open actions 154, 155

jcheney: working on it, please leave it open

Paolo: er, wil get to that, thanks for the reminder

pgroth: action 156 to be discussed in the XML section of the agenda

<tlebo> :-)

Congrats CR

<pgroth> http://www.w3.org/blog/SW/2012/12/12/a-major-release-of-prov/

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

pgroth: we went to CR, this implies a major release of the whole set of docs. now is the time to advertise these. may use blog and web page for this
... encourage people to comment, implement, use

Paolo: I will send to DataONE as I have done in the past
... will send to DBWorld as well

<jun> I can send to hcls list

ivan: will post to sem-web list

<hook> there is also the Federation of Earth Science Information Partners (ESIP)'s semantic web mailing list and the preservation & stewardship mailing list

<hook> sure

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

<tlebo> ok.

<zednik> ok

<dgarijo> ok

pgroth: editors to update the drafts back to "editor's draft" status

<dgarijo> ok

pgroth: please dgarijo check the link to DC
... we've got nice PROV logos

WG Implementations

pgroth: please fill in implementation survey, so we know how we are going to meet our exit criteria

<dgarijo> **linked fixed in the page**

<dgarijo> I will fill in a survey

pgroth: in particular if an impl. builds upon (?) or connect with another impl

<Dong> Southampton will submit reports soon (by the end of 2012), 9 applications in total

<jun> ok, thanks!

<dgarijo> so, internal deadline: First week of January. Got it

pgroth: official deadline end of January, but internally fist week of Jan. would be ideal, so we know where our gaps are

PROV-AQ Reminder

pgroth: a number of issues on the list by GK

<GK> I just joined the call. Will trtyto field any questions.

<pgroth> i think i will do it

GK: (very hard to hear)

<ivan> graham, we do not understand you

<GK> OK. VOIP problems again.

pgroth: (reporting for GK)
... major proposal ew need comments on: we introduced a description of content negotiation -- in spec. provenance services
... this is new to the doc
... also updated def. of prov services description, specifically on whether our use of RDF for service description is appropriate
... also support for SPARQL query endpoints that can answer questions about provenance
... does that require a new link type? (?)
... also provenance pingback -- forward pointers to provenance

<GK> It's not using *provenance* from somewhere else…. it's generating provenance somewhere else...

<GK> … i.e. using the resource, and being able to provide priovenance back to the resource spublisher

<GK> Im thinkl you giot it.

pgroth: please all look a these issues and contribute to the discussion on the list
... hopefully all sorted by 2nd week in Jan

<GK> I also need to follow up some responses from LDP particpants

PROV-Dictionary

<TomDN> http://dvcs.w3.org/hg/prov/raw-file/default/dictionary/prov-dictionary.html

<TomDN> http://dvcs.w3.org/hg/prov/raw-file/default/dictionary/PROV-Dictionary_discussion.txt

TomDN: need to open the txt file above to follow the discussion...
... problems with the constraints and notation the editors did not like
... txt file includes new proposed notation
... problem is that all membership must be in one relation. This won't work for long lists
... propose the hadMember notation to align with Collections. multiple such statements are allowed

<tlebo> It's been a while since this group's made a design decision. Do we still remember how to do this ;-)

TomDN: the proposed change is local to the dictionary doc

<tlebo> KeyValue pairs can be Entities.

smiles: is that really true that this has no effect in prov-n? now the second parameter is no longer an entity

TomDN: yes but that's one of the extensions for dictionary

<pgroth> ack

<GK> Hmmm.. if entities can be key-value pairs, then maybe can align with LDP containers proposal more?

pgroth: if there are no issues with this, it's ok to go ahead with the changes, but give the group an opportunity to review them

<tlebo> +1 on issue 1

<Luc> just stepping in, without having heard the discussion: it may be problematic to have hadMember(c,{k,e}) {k,e} is not an entity, but e is

TomDN: issue 2 is on completeness of dictionaries

<GK> cf. http://www.w3.org/TR/ldp/#linked-data-platform-container

TomDN: old notation (with the 'true" flag) is problematic -- see the comment in txt file, section 2
... proposed / alt 1: add complete attribute to dictionary itself

<MacTed> GK - is there a conflict between LDP containers and what we're discussing (i.e., would what's here break LDP containers)? note that PROV is general case, and LDP is a specific case, so they needn't be in perfect sync; e.g., LDP may be more restrictive

<GK> I'm worried that this might fall foul of RDF monotonicity

TomDN: proposed alt 2: start from EmptyDictionary, then insert
... the result must be complete

<GK> @MacTed - not seeing any breakage, just trying to make sure we're aware and making sure things can be used together. I guess my thinking is that (if it makes sense) use LDP structure as base and focus PROV effort on container-based provenance

<MacTed> GK - I'm not understanding your concern. "RDF monotonicity" meaning?

<tlebo> I'm not sure you'd "be sure" that it's complete in ALTERNATIVE 2... since other derivations could have inserted elements.

<GK> @macted - meaning that it should not be possible to invalidfate anyinference by adding a new RDF statement

<MacTed> GK - "Linked Data Platform" is not parallel to nor core of "Linked Data" nor "RDF". interpretation based on naming is unfortunate.

<tlebo> +1 to "I'm telling you that I think it's closed" as opposed to relying on walking through a derivation to see.

Paolo: does alt 2 really entail completeness?

<GK> i.e. whenever a |= b then a \/ x |= b for any x, where a, b and x are RDF graphs.

pgroth: can we leave both of these in the draft and have people discuss/select?

TomDN: sec. 3 is on constraints

<GK> @macted - agreed, but if it makes sense to re-use it seems that would be a Good Thing.

TomDN: seeking help with the very last constraint

jcheney: conclusion of the rule can be fixed and formalized (each member of d1 is also a member of d2 and vice versa)
... this requires a more expressive logic than what we currently use

<MacTed> GK - I think LDP is too much moving target, and also too much "subset" to be considered for this re-use.

Paolo: last constraint effectively *defines* that provenance of dictionaries is complete

pgroth: next steps: you could solicit a discussion on these issues, and then go for a proper review
... or: we do a draft first, then "discuss by review"

<GK> @macted I more than partly agree. OTOH, don't want to completely ignore what seems to be a significant effort. I was specifically asked to consider LDP views for PROV-AQ stuff (which I know isn't the same thing, but the principle seems applicabl;e).

TomDN: agree on option 1

pgroth: so please start a discussion and then we will appoint reviewers

TomDN: nothing about prov-xml in the doc. are the prov-xml people planning to implement dictionaries? if so they would be best placed to add this part

<tlebo> seems like it's not stable enough to fill out the PROV-XML examples.

pgroth: best to first agree on these issues, add XML examples later

<TomDN> :)

<tlebo> +1 great stuff, @TomDN

Prov-xml

<TomDN> @tlebo, tnx!

pgroth: status update?

zednik: FPWD with good feedback from the group
... still processing the feedback
... need to differentiate the two XML serial. that we have (one native, one for prov-o)
... will add naming conventions to the editor's draft. should be ready very soon

<smiles> Could someone raise an issue for the primer, so I can be clear what is required?

pgroth: any feedback from xml people?

<hook> we also got feedback from Stian on namespaces http://www.w3.org/2011/prov/wiki/ProvXMLNamespaces

zednik: early to tell

pgroth: comments?

<pgroth> prov:Entity

<pgroth> <prov:entity

pgroth: people who looked at the XMl in the primer under Turtle, thought it was for the RDF. that was confusing
... need to clarify

<MacTed> putting an inline comment in the example(s) might be worthwhile...

smiles: problem is there are many examples, it would be messy

MacTed: there is a risk we are creating confusion, can't expect others to be clear about the distinction amongst the XML versions

<tlebo> @smiles, perhaps replace "XML Example (hide all)" with "PROV-XML Example (hide all)"+= link to prov-xml in every title to an example.

pgroth: it's just a matter of clarifying that prov-xml is not prov-o xml
... only have one type of XML serial visible

<tlebo> -.5 to @pgroth 's "show only one"

<smiles> @tlebo Could do, certainly, but I'm not clear if it completely solves the problem

<tlebo> @smiles every bit throughout helps.

zednik: technically it can be easy to add the message to convey the distinction without too much manual effort

<pgroth> that's fair enough

<Curt> Would it be worth adding a sentence in the OVERVIEW saying PROV-XML is not an RDF/XML version of PROV-O (in addition to adding to primer)

zednik: we should present rather than hide the distinction

<MacTed> +1 present distinction, don't hid it. confusion won't only come in *here* -- what happens when PROV-XML is encountered in the wild, and taken for RDF/XML?

<tlebo> +1 to seeing the entire document. That's why PROV-O's examples have full TTL in every one.

Luc: XML examples contain just entities, if we added the context, would it be clear enough indication that it's native XML

<tlebo> (and adding the <xml> bit at the very top)

Luc: i.e., by adding the root elements to the examples

<zednik> +1 to show entire <prov:document> in xml examples

Namespace

<Luc> @smiles: instead of say XML example, can we say PROV-XML example?

pgroth: status update on XML namespace:?

<tlebo> yup, I"m fine with it.

<tlebo> ah, sorry. I thought you were referring to the @xmlns: issue...

pgroth: on merging multiple docs into one ns in XML:?

zednik: need to look at what stian is proposing

<pgroth> http://www.w3.org/ns/prov#

pgroth: need a glossary off the landing page

<MacTed> +1 "say PROV-XML example"

<tlebo> bye!

<SamCoppens> bye!

<dgarijo> bye

<Dong> bye all

<GK> Bye

<zednik> bye

<pgroth> trackbot, end telcon

- DRAFT -

Provenance Working Group Teleconference

13 Dec 2012

Attendees

Contents