Provenance Working Group Teleconference -- 12 Jan 2012

<Luc> are you back?

<trackbot> Date: 12 January 2012

<jun> Yes! Glad to be back! First week back at work!

<Luc> congratulations!

Hi Jun!

<Luc> only a few thousand messages to catch up with ;-)

<jun> @luc: thanks

<jun> @luc, yep:)

<pgroth> scribe?

<satya> I can scribe

<pgroth> Scribe: Daniel

<pgroth> Scribe Daniel Garijo

although I have a bad sound quality today :(

Admin

Paul: as usual, vote on the minutes of last week

<pgroth> Minutes http://www.w3.org/2011/prov/meeting/2012-01-05

<satya> +1

<jcheney> +1

<jun> +1

<tlebo> +1

<khalidbelhajjame> +1

<Curt> +1

<Paolo> +1

<MikeLang> +1

<pgroth> Accepted Minutes January 5 2012

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

Paul: review of actions items
... 1) Action on Paul to write a blog on overview
... about the PROV activities

<pgroth> http://www.w3.org/blog/SW/2012/01/11/feedback-welcome-an-overview-of-the-provenance-prov-family-of-specs/

Paul: Luc had an action item to write a blog post with the diffs
... Satya had an action to look at a number of issues

Satya: I have been working on it

Paul: please sign on the f2f page so I can make the appropriate arrangements
... we need scribes

Prov-AQ changes

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

Paul: gk tried to address a number of issues
... made a review and a bunch of editorial clarifications

<pgroth> http://www.w3.org/2011/prov/track/products/5

Paul: question about xml examples. Do we want them? where should they come in?
... ??? suggested pingback to know when the provenance has been recorded

<pgroth> +q

Paul: paul has to still to review the changes to see if completely agrees with it

<pgroth> prov-dm

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

Paul: update on issues in prov-dm
... most of the issues have been raised/resolved.

Luc: we have to group the issues. 105 is still open.
... will follow up a response

<satya> @Luc, yes I am now reviewing the updates in DM and will respond to these issues soon

Luc: tomorrow morning will start another pass on the document to prepare it for review by the next telecon

<Luc> @satya, thanks

Luc: please satya review the pending issues

satya: ok

paul: are this already in your action?

Luc: no. That action is from several weeks ago

<AndroUser> the "vacant" conference bridge is fighting me

paul: satya, replace the action with a new one.

<pgroth> ACTION: Satya to address issues in http://www.w3.org/2011/prov/wiki/Meetings:Telecon2012.01.05#prov-dm [recorded in http://www.w3.org/2012/01/12-prov-minutes.html#action01]

<trackbot> Created ACTION-52 - Address issues in http://www.w3.org/2011/prov/wiki/Meetings:Telecon2012.01.05#prov-dm [on Satya Sahoo - due 2012-01-19].

prov-o

<satya> Meeting minutes from last PROV-O call: http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology_Meeting_2012-01-09

satya: dgarijo has created a list of issues, and we have been resolving them
... still have 3 main bullets to complete the list.
... we have been recording the changes to make in the doc
... some issues have been raised.

<Luc> can you indicate what requires discussion with dm?

satya: you'll see the changes to the html doc soon (next telecon)

Luc: can you identify what are the issues to be discused within dm?

<satya> Meeting minutes PROV_O: http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology_Meeting_2011-12-29

<tlebo> http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology_Meeting_2012-01-04#Did_not_get_to_in_this_telecon ?

satya: some concerns about making wasStartedBy and wasEndedBy subclasses of wasAssociatedWith

Stephan: some classes seem to be modeling things with different semantics. Also actedOnBehalfOf

luc: some of the issues could be raised as issues against prov-o
... none of the editors will be at f2f

<khalidbelhajjame> I ll be there

<tlebo> I'll be calling in to F2F2.

satya: some of the authors are going to attend: Daniel, Stian, Khalid

<AndroUser> not tim?

paul: so, in summary, you will be raising issues against dm soon.

satya: yes

<jcheney> www.w3.org/2011/prov/wiki/FormalSemanticsStrawman

Prov-Semantics

jcheney: I've been updating what's there
... answered some emails
... diference between the 3 level and 2 level ???

<jcheney> http://www.w3.org/2011/prov/wiki/FormalSemanticsStrawman#Objects_vs._Things_vs._entities

jcheney: what the alternatives are about this issue

luc: this could be raised as an issue against the semantics

jcheney: will do that

satya: Ithink the distinction between entities and real world entities is very important

<stian> +1 to satua

satya: the real world things have no relevance in our context

<Paolo> @satya: not relevant for the language, but important distinction for the semantics!

<Paolo> so "things" are not part of the description, but we are talking about the semantics here, i.e., the interpretaion of the language constructs

luc: From the beggining we made a distinction between the record and the real world thing. Thus it is part of dm

<satya> @Paolo: In our context that is any computer science application - there are no real world things

@satya: what about the provenance of Mona Lisa? That is a real world thing..

Paolo: we're not talking about provenance within a specific system.

<stian> @dharijo no, you are talking about a characterisaton

<satya> @Daniel: No, the "Mona Lisa" in any application is a representation/abstraction of the real world thing - the real world thing is never part of any computer science application

paul: maybe we can pick this at the end of the call

<stian> is it the painting? the model?

@stian, satya: ok

<stian> (have to go now)

luc: will try to catch up with the tracker.

jcheney: we have to identify where are we going to find this at the time of the f2f (next 2 weeks)
... alternate of, specialization of semantics.

paul: when do you like people to read the document?

jcheney: people can look at it now and provide feedback

<Paolo> sorry I have to go...

<satya> I will review

pgroth: volunteers?

<pgroth> \

pgroth: paul, satya are volunteers.

Accounts in Prov

tlebo: I haven't had a chance to answer all the people yet

<khalidbelhajjame> +q

tlebo: if anibody has something right now it will be ok

khalid: how entities are differentiated in different accounts
... ?
... it's more a practical point of view.

tim: TRIG syntax is a bit confusing in those examples.
... how can we have this scoped entities without the dcterms:identifiers.

khalid: so you think there could be problems having different entities scoped in different accounts

<satya> @Tim: +1, also having scoped identifiers (aka URI) is contrary to RDF semantics and global scope of URIs

khalid: how do I identify an entity across different accounts.

Tim. the same URI is mentioned in both named graphs.

scribe: there is no requirement that the inner accounts have to be mentioned in the outer accounts

khalid: thanks

paul: are you having problems with ids in dm?

tim: maybe I'm misreading ids in dm

luc: your interpretation is ok
... I'm not sure that we have the same understanding of record

tim: in the rdf world a reocrd is a triple or an rdf graph.

<satya> @Luc, what are the columns of these tables - Entity, Agent, wasGeneratedBy?

luc: that would be rooted in a specific subject.
... if we have an entity Luc in Boston type person type entity, everything would be rooted from that.

MacTed: a record in my mind is a single row in a database. In the rdf world is a triple

luc: a record in dm is a set of triples in the rdf context

<zednik> record

<zednik> noun |ˈrekərd|

<zednik> 1 a thing constituting a piece of evidence about the past, esp. an account of an act or occurrence kept in writing or some other permanent form: identification was made through dental records | a record of meter readings.

MacTed: so you're changing the understanding of record that is different in many areas.
... I keep going in circles
... the common terms are not used according to their common meanings.

luc: I disagree
... nobody has suggested an alternative to this terms

<zednik> database terminology is not common term usage

satya: if an entity is mapped to a table, what would be the columns of the table?

luc: activity, sart, end, etc.
... (example)

<Luc> http://dvcs.w3.org/hg/prov/raw-file/default/model/overview.png

<tlebo> it's sounding like "expression" _was_ a better term (the original "record"). What motivated the rename?

luc: satya, how are you suggesting to express that?

<MacTed> :LucInBoston :wasWearing :whiteShirt

satya: Luc in Boston is an entity

<tlebo> entity(:luc_in_boston, [:shirt_color "white"])

satya: discussion about the Luc in Boston entity.

<Luc> entity(:luc_in_boston, [:shirt_color "white", :name="Luc", :type="Person"])

<Luc> entity(:luc_in_boston, [:shirt_color="white", :name="Luc", :type="Person"])

satya: I was trying to understand what an entity record mean. Luc in Boston has type person would be an entity record

Luc: I've extended what tim just posted
... you would map this to a series of triples
... entity record seems to map to a set of triples

satya: this is an example of relational db to rdf

luc: I'm not reinventing the world

<Luc> http://dvcs.w3.org/hg/prov/raw-file/default/model/overview.png

luc: I've posted the diagram of the document. That is how it would be recorded in a relational world

satya: it is an assertion about luc in boston. It is an important distinction to make

<tlebo> I'm wondering what motivation we had to rename "expression" to "record".

<tlebo> +1 to satya pointing out the confusion of identifying the record or the characterized thing

<tlebo> @luc, thanks.

luc: I'd like to explain what paulo and paul have been discussing. There are some inconsistencies in dm and we're trying to resolve them.

<tlebo> luc: too much "language association" and was inappropriate.

<MacTed> so, "entity record" is a collection of (one or more) assertions about an entity...

<MacTed> (and an "entity record" is an entity in its own right, with assertions about it, etc.)

<MacTed> some of the assertions about the "entity record" comprise the provenance of that record

luc: every record has an id

<GK> +1 to Satya too (the "record" is artifact of ASN; shouldn't have representation in RDF translation)

satya: no, every entity has an id

<GK> What Luc is now describing is artifact of "the relational view" i.e. an implementation detail for *some* implementations.

paul: let luc describe the problem, and we try to solve it offline

luc: coming back what satya was saying
... luc in boston in my view is not the key of the record.

@tlebo: thanks, sorry.

<Luc> entity(:luc_in_boston, [:shirt_color="black", :name="Luc", :type="Person"])

<Luc> entity id + account id = natural key for entity record

<satya> exactly - so luc_in_boston is key for Entity "Luc in Boston"

scribe: luc in boston could have different color of tshirts, but it would be the same entity

<satya> not the record

<tlebo> luc: white and black shirt on same :luc_in_boston

<satya> @Luc, then we need a distinct identifier for the record itself

<Luc> entity_reocrd_id001: entity(:luc_in_boston, [:shirt_color="black", :name="Luc", :type="Person"])

scribe: an approach to this is to have an entity record Id that would be the key for that record. Now I would need an extra column

<satya> @Luc: ok

scribe: if we do that, that's great.
...

<GK> @satya - isn't the record its own identifier?

scribe: we can simplify a lot of the text

<tlebo> entity(:luc_in_boston_on_Monday, [shirt_color="black", location=Boston, name="luc", specializationOf=:luc_in_boston_in_July])

<MacTed> "Named Graph" :-)

<tlebo> entity(:luc_in_boston_on_Tuesday, [shirt_color="white", location=Boston, name="luc", specializationOf=:luc_in_boston_in_July])

<Luc> entity id + account id = natural key for entity record

scribe: but from the rdf perspective may not be nice. It would imply new ids

<satya> @Luc: the natural key for entity record is something different from key for entity

<tlebo> @luc, but that throws the benefits of URIs out the window (and violates AWWW)

<satya> @GK: I didn't understand

<GK> I think we are trying to turn ASn "records" into implementation artifacts, when they are explicitly an *abstract* syntax for talking about provenance assertions.

<MacTed> G-box would give the ontology of the accounts (i.e., the schema of the "entity records")

<MacTed> G-snaps would be the account ("entity record") instances

<MacTed> G-texts are the serializations of those instances

paul: the issue is clear.

tim: trying to respond to Luc about naming the resource within the account.

luc: luc in Boston is the name of the entity.

MacTed: so the entity could have 1 million URIs, identifiers, names
... and be referring to the same thing

<GK> Why does ASN use URIs anyway?

MacTed: problem on discovering other descriptions of the same entity the first time that I'm going to describe it. How do I know that there are others?

luc: the uri luc in boston is not enough to identify the records
... that is why you need to know which account belongs to

<tlebo> @luc, then you mistakenly named luc in account 2.

<Luc> @tlebo, why?

<tlebo> @luc, you knew that they are different, but named them the same thing.

satya: adding the acocunt id + the record does not make it an ? entity record?

<Luc> @tlebo, no, it's intentional, I am giving two hypothesis about what luc did

luc: raising issues might be the best thing

<pgroth> trackbot, end telecon

- DRAFT -

Provenance Working Group Teleconference

12 Jan 2012

Attendees

Contents

Admin

Prov-AQ changes

prov-o

Prov-Semantics

Accounts in Prov

Summary of Action Items

Scribe.perl diagnostic output