Provenance Incubator Group Teleconference -- 19 Nov 2010

<trackbot> Date: 19 November 2010

trackbot, start telcon

<trackbot> Meeting: Provenance Incubator Group Teleconference

<trackbot> Date: 19 November 2010

chair Yolanda Gil

<scribe> Scribe: Luc Moreau

<scribe> ScribeNick: Luc

<YolandaGil> thanks Luc!!

<Paolo> ??P1 is me

<YolandaGil> Luc: do you really want to lead the discussion and scribe at the same time?

<pgroth> I'll lead the discussion

who's on the phone

Paul: 1. try to agree on set of deliverables
... 2. Discuss the scope of the language
... three common deliverables in the three proposed charters
... primer, best practice cookbook, inter operability guideline
... is there agreement?

simon: what is the diff between primer and best practice?

Paul: primer: educational/outreach purpose

<Paolo> no objections on D3-D6

Paul: best practice: how does it link to dublin core, creative commons, and other ontologies out there

<pgroth> Best Practice Cookbook (W3C Note). This document includes a limited set of best practice profiles that link with other relevant models, such as Dublin Core provenance-related concepts, licensing in Creative Commons, and the OpenId identity mechanism for people.

<Paolo> (but where is D4?)

<pgroth> XG Primer (W3C Note) Educational/outreach material aimed at users of provenance.

Paul: it looks people are happy with these deliverables. Let's discuss the others.
... The notion of abstract model
... ... needs to be clarified

<pgroth> is document includes includes the data model and permitted inferences.

<pgroth> his document includes includes the data model and permitted inferences using OWL and Semantic Web technologies.

Paul: data model and set of permitted inferences

<pgroth> his document includes the syntax and the conceptual model

Paul: data model and inferences in semantic technolgies
... abstract model is definition in english + nice pictures, without being tied to any technology

Satya: we need a syntax to express what we're are proposing, to avoid ambiguity

Paulo: there are languages (formal) to use and express this (doesn't have to be owl).

Paolo, which one do you think about?

Deborah: supports the same statement
... just english, i can't leave with!

Paulo: why need of something less formal?

Paul: why xml?

<ssahoo2> Without first deciding on the syntax to represent our abstract model, we will not know whether what we express in English can be expressed in a given formal language

<Deborah> Does that mean that we might generated 3 things: 1 - a description in English, 2 - a description in some graphical notation (is there an example? is cmap an example), 3 - a description in a formal language (e.g., OWL, but other examples are SCL)

Luc: important to be able to express ideas indpendently of a technology. It is complementary to a schema/ontology.

<SamCoppens> +1 for a data dictionary + formalised representations in line with it (XML, OWL, ...)

fror me, two things: 1 english+illusration 2 owl ontology

Paul, can you repeat?

Paul: 1. Reference model (or data dictionary): english language definition

2. Formal definition of model

Paulo: for communication purpose, it's nice to have english description

<ssahoo2> I don't think is whether to have 1 or 2 - its the ordering I am concerned about

<Deborah> I suggest we use a phrase other than reference model for the Natural Language description plus illustration)

Paulo: ... so that people who are not familiar with formal notation can still understand the model

deb, do you have a suggestion?

<pgroth> ack

Paul: do we have consensus for: Reference model in NL and formal model?

Satya: definition in English should be compatible with what we express formally, since each notation may have its limitation

<Deborah> For the moment, I am happy with your earlier description; two things : 1. English language description with some graphical illustrations and 2. a formal description possibly using OWL.

<Deborah> +q

<Paolo> +1 for the ref / formal model distinction

+1 with Satya

<ssahoo2> conceptual model

<jun> +1 yes, it's more like a conceptual model to me

Deborah and Paul: we need a proper term to describe the reference model + NL + illustration

<ssahoo2> yes

why not name them Deliverable 1 and D2?

<Paolo> somebody very noisy

Paul: the other deliverable for which we may have disagrement, is formal semantics

<ssahoo2> +q

Luc: it should be optional, we may not have critical mass to do that work

Satya: semantics important,
... non optional

Deborah: non optional
... we don't want different interpretation. It's important for inter-operability

<DGarijo> +1 to what deborah said.

Paolo: obligation to be as formal as possible. We need to find the people to do that.

<ssahoo2> I agree with Paolo that both 1.2 and 2.1 are overlapping

Paul: it seems we all want formal semantics deliverable.

<jun> the term mapping exercise has shown it's vital to have a clear semantics

<SamCoppens> +1 for semantics

Paul: agreement on a required formal semantics deliverable

i droped from the call, can someone minute, I am dialing in

<DGarijo> Paul: accesing and delivering provenance

<Paolo> which Del is that in the proposed charter?

<DGarijo> Paul: we agree we need the deliverable

<Paolo> access/delier provenance

back, thanks

<YolandaGil> Paul: we have agreement that a deliverable will be semantics of the model (not optional)

<ssahoo2> is it accessing and delivering or accessing and querying?

<Paolo> is delivery a serialisation model?

Paul: should we have an xml serialization? what serialization?

<Paolo> @Satya serialization not query

<ssahoo2> ok

<ssahoo2> yes

<DGarijo> yes

Paul: agreement we want the model in rdf and owl

<SamCoppens> +1 for owl

<Deborah> +1 to RDF/OWL

Paul: what about an xml version?

<smiles> +1 for OWL

<Paolo> +1 to RDF/OWL

+1 for xml, for reaching out beyond semantic web community

<DGarijo> +1 to RDF/OWL

<ssahoo2> can we consider RDF/XML as XML?

i don't think so

<Paolo> @satya -- not really

Paul: the goal is too have an XML that looks like real XML

<ssahoo2> for provenance exchange language specific XML serialization we will need to define our schema

Simon: given time pressure, shouldn't we go for 1 serialization?

<Paolo> @Luc can we get members of the XML data community involved to contribute to that part

<ssahoo2> I think XML serialization is needed, but we should be mindful that we will need to define a schema (which may end up looking like RDF schema)

Luc: xml serialization is low hanging fruit, which we can get out of the door quickly

Paulo: waste of time to go with XML, + lack of semantics

why is it that owl2 has got xml serialization and rdf serialization?

Paul: we have agreement for rdf serialization. Do we want xml serialization?

Jim, do it as part of the best practice/primer guide?

<ssahoo2> +1 for XML + schema (suspect that the schema will end up as RDF/XML :))

<Paolo> @satya I suspect the schema will be nothing like RDF/XML :-) just like OPM/XML is not but I support the idea of a mapping to an XML schema

<SamCoppens> Provenance will end up using XML and OWL instances, so those two must be interchangeable, +1 for XML

Paul: the xml schema is a separate deliverable, developed in the context of best practice
... OK, we seem to have agreement
... the formal model, what formalism to use? should do it offline, or now?

<Paolo> agree that formal and "narrative" models should proceed in parallel

<Deborah> +1 OWL

<DGarijo> +1 to OWL

<ssahoo2> +1

<SamCoppens> +1

Luc: reasoning may be challenging to express in owl, we should be open to consider other semantic technologies, e..g rule based

Paolo: how do we know OWL is expressive enough?

why not be open, and suggest Semantic Web technologies, e.g. OWL, ...

<DGarijo> so first decide the inferences and then the language?

Satya: if we adopt OWL, we can leverage the tools out there

<Deborah> What if we say we will deliver an encoding in a formal language, most likely beginning with OWL

BTW, this is the very reason why a NL language description is crucial! Technologies may be limiting.

<Deborah> +1

<ssahoo2> +1

<smiles> +1

Paul: formalization will be in using Web Semantic technologies, beginning with OWL.

Paolo: we shouldn't use the word technologies

<DGarijo> swrl rules could also help in the inferences...

<Paolo> @daniel good point

Paul: last thing to decide is order
... do we do english first, formal second, or vice versa??
... alternatively, simultaneously?

<ssahoo2> +1 for Jim

Jim: yes, working them in parallel makes sense

<Deborah> +1 to simultaneous

<smiles> +1

<Paolo> +1

<DGarijo> looks fine to me too

<SamCoppens> 1+ for parallel

<Paolo> minor note: in Satya's version D4 has disappeared but it is still in the timetable on last page

Paul: good, we have some consensus!!!!!!!!!!!!!!!!!
... hhow precise should the charter be about what the WG will do?

<ssahoo2> sorry typo - I think since we are going for RDF/OWL we can just leverage SPARQL

Paul: first proposal is to identify a set of terms to work with, terms defined in various vocabularies out there
... second proposal is to leave this scoping exercise to the WG

<ssahoo2> I can too

<jun> me too

<Paolo> not sure this is fair to people who have to go

<DGarijo> me too

<Deborah> i have to go to a meeting at noon. i can be a few minutes late but that is it. sorry

<ssahoo2> @Paolo has a point

<ssahoo2> can we have a one-off meeting early next week?

Paul: proposal 1 was in fact a suggestion by TBL
... ... list the terms (and not just from OPM!)

Jim: objects against the word 'term'. Would be better to talk about concepts.

Paul: OK, a set of concepts

<ssahoo2> I dont think we can do it now - so should be part of WG (second option)

<ssahoo2> @Paulo: exactly, we need time to discuss and come to consensus

Paulo: need to go through examples to identify terms

<ssahoo2> +1 for procedure

Paul: agree on a procedure to list concepts and examples

<ssahoo2> yes

Paul: procedure is: list concept, and identify an example
... this procedure is what we do *now* as an incubator, as an input to the charter

Yolanda: can we use our scenarios?

Paulo: the scenarios may be too complicated

Jim: lots of use cases illustrate the concepts.

Paul: agreement is to list concepts with small examples, drawing on scenarios and use cases
... what is the timeline?

Yolanda: hard deadline is Nov 30th

<ssahoo2> with Thanksgiving holidays it seems daunting

Paul: can we have this by tuesday or wednesday next week?

<Paolo> no thanksgiving in EU so it's fine :-)

<SamCoppens> +1

<jun> do you want a list from all possible vocabularies?

Yolanda: we don't have much choice

<Paolo> Tue/Wed is fine

Yolanda: we should be mindful that we shouldn't blow up the scope, otherwise 2 years will be too short
... we need to prioritize (mandatory/optional)

Paul: can be from any vocabulary

<ssahoo2> I would also suggest members to look at list of terms on the mapping wiki

Paul: ACTION: send a note later!

trackbot, end telcon

- DRAFT -

Provenance Incubator Group Teleconference

19 Nov 2010

Attendees

Contents

Summary of Action Items

Scribe.perl diagnostic output