Provenance Working Group Teleconference -- 20 Oct 2011

<trackbot> Date: 20 October 2011

<pgroth> Scribe: Stephen Cresswell

<pgroth> hey graham, just out of curiosity why do you also have two irc names?

<GK1> @paul I'm using two machines .. let's me track the IRC while digging around for dicuments, etc. :)

<pgroth> cool :-)

<stain> hi, guys

<GK> Hi Stian

<stain> note to self: always pause between the seven keys

<stain> @kai does "probably" allow others to override..?

<kai> probably ;-)

<kai> Only zakim knows

Admin

<pgroth> PROPOSED to accept the minutes of Oct 13 telecon

<pgroth> http://www.w3.org/2011/prov/meeting/2011-10-13

<dgarijo> +1

<Curt> +1

<kai> +1

<tlebo> 0 (did not attend)

<jcheney> 0 (dna)

<stain> 0 (not read them yet)

<YolandaGil> 0

<pgroth> ACCEPTED Minutes from last week

<GK> 0 (not present)

pgroth: reviewing action items

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

pgroth: Satya's action item on all items from data model in ontology .. we'll come back to it
... scribes needed ...

PROV-DM FPWD

<pgroth> http://www.w3.org/blog/SW/2011/10/18/first-draft-of-a-provenance-data-model-published/

<dgarijo> Khalid sends regrets

pgroth: there's a blog post from Ivan

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

pgroth: is the official link to use

<pgroth> Twitter hashtag: #provwg

pgroth: we want to encourage people to talk about on their blogs etc.
... to get multiple perspectives

<pgroth> +q

<dgarijo> +q

<satya> @Paul: Was there a question for me? Sorry I joined a bit late

<pgroth> no satya

dgarijo: comments said that model was a bit complex, how are we going to make it more simple

pgroth: a number of ways

<GK> The comment address should be in the draft

<dgarijo> maybe more examples? The comments also wanted to see the RDF serialization (prov-o).

<Luc> they can subscribe to the mailing list

where should the comments go?

pgroth: to the mailing list

<Luc> instructions are at the beginning of document

GK: Do they normally have a separate mailing list?

<GK> Doc requests comments to public-prov-wg@w3.org

<Luc> ... it's standard text from w3c

<Zakim> GK, you wanted to note it may be the /presentation/ is seen as complex

<satya> @GK: agree

<stain> @GK +1

GK: It may be not the model that is complex, or is it just the presentation of the model that is complex?

<jcheney> +q

<dgarijo> @GK yes, maybe I didn't express myself correctly, sorry.

<stain> no, not publiclally subscribable. I was redirected to password-protected http://www.w3.org/Member/Mail/

Luc: We have to explore presentation first, and then maybe think about why the model is complex, but maybe it needs to be complex
... thiinking about starting the data model with the most common relations

<satya> @Luc: +1 we should concentrate on presentation but not try to modify the model itself to make it more readable

<stain> @GK just needs some bass

<stain> > This is a response to a message apparently sent from your address to

<paolo> +1 for a two-tier dissemination strategy

<stain> > public-prov-wg@w3.org: Your message has NOT been distributed to the list

<stain> so it does not work to email from the outside either

jcheney: Paolo was giving view of this at conference, he said it may be good idea to start with a subset of ideas that are familiar through OPM etc.
... (gone quiet)
... (I didn't hear anything)

<stain> ah, I just need to say it's allowed in the publication.. then it should be fine to post for outsiders, yes

Paolo: Make distinction between core concepts and newer things
... people ask how it is different from OPM ... make distinction between top tier and second tier

<Zakim> GK, you wanted to say that I've been taking a shot at entioty/resource stuff

<YolandaGil> +1

GK: I've been starting to draft something about the issue of relationship between resources and entities, from developer perspective,
... intend to contribute

<tlebo> FROM @MacTed ASK WHERE { [ foaf:nick "Tall Ted" ] foaf:holdsAccount [ foaf:accountName "MacTed" ] } . }

Yolanda: Was trying extract core ideas to write primer document.
... some of the examples are hard to relate to

<GK> Slightly related to this discussion: http://lists.w3.org/Archives/Public/public-prov-wg/2011Oct/0156.html

<GK> Yolanda +1

Yolanda: some of the the definitions , or how terms are used elsewhere in document, are confusing

<stain> @YolandaGil +1

<tlebo> +1 to stop making contrived examples when there are many real examples to handle

<satya> @Yolanda +1

Yolanda: may less contrived, more natural examples, more like the way we would use provenance, would help

<tlebo> attempt of a list http://www.w3.org/2011/prov/wiki/PROV_OWL_ontology_component_examples

<dgarijo> @Yolanda +1

Yolanda: perhaps end-to-end example using linked data

pgroth: Would help to get a reflection of the data model in RDF would help

<GK> @pgroth: +1 helps to see examples in RDF, but these can still look complicated

pgroth: we should look at smaller things to help people understand what the data model says

<satya> @Paul +1 and maybe with an intuitive example will help

pgroth: it's nice to have the RDF as well as abstract syntax

<dgarijo> we have some examples in RDF in the ontology document.

<tlebo> RDF/XML :-(

satya: linked data, bioinformatics, sensor data, has lots of examples
... maybe better than filesystem examples

pgroth: We have the example that we all agreed on previously ... data journalism example

<satya> @Paul: I guess I am saying a non-computer science example may help?

<Luc> we have always said this example was a placeholder ... if someone has a better one, than let's use it. The data journalism is too long for prov-dm document.

<GK> +1 (need examples that demonstrate simple ideas)

pgroth: concentrate on some simple things that everyone would need to say e.g. authorship, quote relation

Connection Task Force

<tlebo> += http://www.w3.org/2011/prov/wiki/PROV_OWL_ontology_component_examples#Who_is_the_author_of_a_document

<satya> We can revisit some of the example scenarios from the PROV XG also

<tlebo> @satya, pointer to XG's list?

<dgarijo> @satya: well, the first one wasn't the data journalism example?

kai: We have telecon discussing what we will do with connection TF,
... so far we have informal rep ... more or less finished
... it should be a living document
... eric and kai will be a contact
... What will we do next?

<dgarijo> @tlebo: http://www.w3.org/2005/Incubator/prov/XGR-prov-20101214/#Analysis_of_The_State_of_the_Art

kai: one thing is to identify mailing lists

<tlebo> thx

kai: we don't want to be seen as the people who actually communicate with all these connections

<satya> @Daniel: agree but maybe we should avoid CS jargon - since many of our targeted users are non-CS

kai: it would just add another step to communications
... We brainstormed on what else we could provide

<dgarijo> @tlebo: 5.3.1, 5.3.2 and 5.3.3 are the 3 3 scenario.

kai: We thought about organising additional telecons
... we other groups, e.g. creative commons
... What does the group think?

<dgarijo> @Satya: what does CS mean?

<satya> @Tim: http://www.w3.org/2005/Incubator/prov/wiki/Analysis_of_Disease_Outbreak_Scenario

pgroth: Still a lot to do in engaging with other groups

<satya> @daniel: computer science

pgroth: e.g. kai involved with DC group

<dgarijo> @satya: thx!

Luc: In charter we have Best Practive deliverable
... would involve technical work on how integrate e.g. creative commons work

<YolandaGil> +q

<Zakim> Luc, you wanted to mention the best practice deliverable

kai: That's sort of thing what we might discuss on new telcons
... What can be our role where we are not bridge persons?

<paolo> have to leave, apologies

yolanda: There are not many people on connection TF, and it is too daunting to look at technical integration at all these areas ourselves
... we need to set up discussions in these separate areas
... telcons would be vehicle to get more people participating, and the outcome from calls would move us towards best practice deliverables

pgroth: seems like a good way forwards

yolanda: Maybe people will be scared off by prov-dm document, and maybe we should hold off until we have more accessible documents

pgroth: seems reasonable to wait for primer etc.

<Luc> we should use the w3c teleconference system for those calls, and we need to book them ahead

<Luc> +1 for January

yolanda: maybe we can schedule for later ... e.g. Decemeber, January

PROV-O

pgroth: January seems best.

<tlebo> http://www.w3.org/2011/prov/wiki/Reflections_from_RDF-WG_F2F2 : one good (3 terms), two bad (conflating GraphContainer and Graph; sd:name doesn't identify), one opportunity (reconciling SPARQL-WG, RDF-WG, PROV-WG)

tlebo: Some feedback on F2F (named graphs)
... they have resolved to distinguish

graph containers and graph serialisations

<tlebo> http://www.w3.org/2011/prov/wiki/Using_named_graphs_to_model_Accounts

scribe: the clear distinction will help proposal
... they have some problem with the SPARQL WG
... vocabularies used to identify graph doesn't identify graph container
... needs to be solved in SPARQL WG

pgroth: Any impression on whether we just have to wait?

tlebo: We need to be more proactive than that

<satya> @Tim: thanks! The distinction between g-snap and g-box seems to be special importance to this WG

<satya> @tim: agree

tlebo: we need to interact more to make sure the clear distinction is established and maintained

GK: Concerned that we become dependent on what SPARQL WG say

tlebo: Problem is that they have established RDF vocab to talk about endpoints, graphs etc., and they fail to make distinction

<Luc> there was a suggestion by Sandro to express the data journalism example, and trying to use some form of name graph, and learn from that

GK: Their problem or ours?

tlebo: Ours. They could continue to ignore it and they would meet their aims.

Update PROV-O

tlebo: Discussion on named graphs for accounts is stalled by these problems.

<dgarijo> hmm, if we cannot use named graphs as accounts then we will have to include "accounts" on the ontology.

<dgarijo> once again.

satya: Luc joined ontology call and had suggestions before release of documents.
... on data model, it might make sense to withhold prov-o until readablity of dm doc is improved

<stain> @Luc +1 - let's do an agile first version

<stain> with lots of bugs :)

Luc: Would be worried to dealy prov-o document, we need serialisation, for primer etc.

satya: we can go ahead and release

<jcheney> +q

<tlebo> PROV-O is not the RDF serialization?

<Zakim> GK, you wanted to suggest an approach to simplifying presentation of the DM might be via the ontology

satya: to clarify, prov-o is not the RDF serialisation

<tlebo> so, PROV-O is RDF serialization + axioms?

GK: Maybe leave data model as it is, but look at ways through ontology and RDF representation, to make the simple things easy to say.

satya: Agree. Think we have covered the mapping of all the terms in data model.

<dgarijo> @GK: Paul already proposed some shortcut fucntions

GK: Shortcuts may provide a less scary way to present examples

<stain> http://www.w3.org/TR/prov-dm/#prov-dm-extensions

<dgarijo> http://www.w3.org/TR/2011/WD-prov-dm-20111018/#common-relations

satya: Are we considering these things to be part of core data model or as extensions?

<GK> My point was that the ontology could contain things not in the data model, or easier structures to represent DM structures

pgroth: I consider them part of the core

<GK> .. i.e. not necessarily 1:1 correspondence between DM and O

jcheney: Data model uses abstract syntax, ontology uses RDF, but describes constraints and specialisations

<GK> I think the ontology effectively *does* define RDGF serialization

<pgroth> the RDF falls out of the Ontology

<stain> but not with any constraints of formats or implicit/explicit etc.

jcheney: but doesn't describe mapping to ontology

<stain> I assume PAQ should come with some minimum serialisation expectations

<GK> @pgroth yes that's what I meant to say :)

<stain> so you could use PROV-O in Manchester Syntax if you like, but don't expect too many applications to understand it

<MacTed> I don't understand "RDF serialization" nor "XML serialization" in this context. "RDF/XML serialization", yes. or Turtle, N-Triples, etc.

satya: We tried to model DM classes and provide definitions. What is mssing?

<stain> XML serialisation CAN be a (restricted) RDF/XML serialisation

<GK> ACtually it's 2-stage: model -> abstract RDDF (ontology does that), then given that RDF-syntax gives RDF/XML.

jcheney: There's a deliverable about serialisation, is that intended to be serialisation of the of the ontology, or the mapping from the DM to the ontology?

satya: We will add some text on that.

Luc: On issue of time, there don't seem to be any time examples

<dgarijo> we talked about reusing some time ontologies.

Luc: e.g. workflow example, can we have time in there?

satya: We will add e.g. start and stop time of processes.

<dgarijo> like : http://www.w3.org/TR/owl-time/

<tlebo> illustration of owl time: https://github.com/timrdf/csv2rdf4lod-automation/raw/master/doc/ontology-diagrams/owl-time.pdf

<stain> http://www.w3.org/TR/prov-dm/#prov-dm-extensions I mean

<stain> if you are required to understand the extensions if you are "PROV-DM compliant"

stain: Prov-DM extensions, are those something that we are required to understand?

<stain> or if it is optional, so that although PROV-O should have these terms, you don't need to understand it to be PROV-O compliant

Luc: they are part of the data model

<dgarijo> @Satya: entities?

satya: What are the domain and range of the relations?

<GK> ?

Luc: entities

pgroth: It would be good to reflect everything in the DM into the ontology
... not necessarily all the contraints

<tlebo> RDF examples for each construct are in the repository: http://dvcs.w3.org/hg/prov/file/tip/ontology/components

<tlebo> not complete, not correct, but there :-)

<GK> +1 need the vocabulary soonest, not nessecarily constraints

satya: Primary aim should be to get all the terms modelled,
... but if you don't define all the domain, range etc.
... people have problems creating the RDF

<GK> +1 domain and range are helpful for generating RDF

pgroth: Domain and range are mostly there anyway, the hierachies are not so important at this stage.

satya: Agree, but OWL community won't like it.

<stain> example of XSD which happen to produce RDF/XML: https://github.com/myGrid/scufl2/blob/master/scufl2-rdfxml/src/main/resources/uk/org/taverna/scufl2/rdfxml/xsd/scufl2.xsd produces https://github.com/myGrid/scufl2/blob/master/scufl2-rdfxml/src/test/resources/uk/org/taverna/scufl2/rdfxml/example/workflowBundle.rdf

Discussion on Entity Attributes

<stain> .. but you get strange double-nesting due to the property-class nature of RDF/XML

<GK> http://lists.w3.org/Archives/Public/public-prov-wg/2011Oct/0142.html

GK: Link is to one of most recent messages,
... discussion between GK and Jim has converged
... to having attributes as part of characterisation
... to aid interoperability
... Also agreed we don't have to distinguish between characterising and non-characterising attributes

<tlebo> I haven't read the most recent emails on this, but the last time we talked about this, "characterizing attributes" were trying to reinvent owl.

<satya> @tim: +1

<tlebo> (sorry, call also dropped)

<stain> should we do a proposal and vote?

Luc: Didn't mention whether a given attr has fixed value for some interval

<tlebo> all attributes are fixed in an entity, no?

GK: Any attr for entity is fixed for entity in what interval that entity exists.
... Argument for interop came from jim.

<Zakim> Luc, you wanted to ask if you still consider that attributes still have a given value for some interval

<dgarijo> gotta go, sry.

GK: looking at provenance challenge, the attrs were introduced to enable conversion of information between different formats

<kai> sorry, have to leave timely

<tlebo> what is the brewing proposal we may vote on?

GK: the approach I was suggested could be seen as a dual to that

Luc: We will try to get that articulated so we can make the case in the doc.
... are there aspects of the document which conflict with what you agreed with Jim?

GK: Will clarify this.

satya: What GK said is exactly what frames and slots do, and that carries over to OWL.

<tlebo> just as PROV is avoiding the Time and Location discussions, it should also avoid being a schema language.

satya: To make it clear, it would help you don't explicitly carry around attributes of entitty to be able to define it properly, that is done by typing information

<pgroth> trackbot, end telcon

- DRAFT -

Provenance Working Group Teleconference

20 Oct 2011

Attendees

Contents