Media Annotations Working Group Teleconference

22 Sep 2009


See also: IRC log


+, tmichel, pchampin, tobias, Hui, florian, Felix, raphael, Chris, wonsuk
Pierre-Antoine Champin




<trackbot> Date: 22 September 2009

<scribe> Scribe: Pierre-Antoine Champin

<scribe> scribenick: pchampin

<raphael> scribenick: pchampin

Thierry will scribe next week.

<fsasaki> also regrets for next week by Felix (at a workshop)

Next meeting will be on 2009-09-29


Florian: Chris has been updating Wonsuk's proposol with WebIDL
... but may have had some problems with write access to the document.
... Chris had problem with defining the datatypes.
... Should we use "easy" WebIDL types, or define complex datatypes.

Chris: still have to decide whether return types should be structured or not.
... e.g. how do we identify persons or organizations (getCreator).

Joakim: make a proposal on which we can vote.

Chris: we have not reached a consensus on a concrete solution yet.

ACTION-155 ongoing

Joakim: late for the draft document. Can the editors make a proposal by the end of the week?

Chris: we can do that.

Joakim: please update us on the particular issues you encounter.
... Felix, do you have an opinion on the question of datatypes?

Felix: I would prefer easy WebIDL types.

<fsasaki> Felix: but have to re-think this again properly

Joakim: we should take a decision soon, even if we need to backtrack later.
... I have to go. Florian and Chris can discuss those issues with the rest of the group.

Chris: we now have a WedIDL representation of the models and properties.
... Regarding the 'contributor' property, how do we represent persons or organizations.
... For the moment: a string, which can be a URI.
... Any better idea?

Pierre-Antoine: applications should be able to decide whether the string is intended to be a URI or not.

Chris: in WebIDL, we can fix the meaning of the string (using comments)

Pierre-Antoine: but in many underlying formats, you won't always have a URI, so you should be able to provide a literal string

Chris: depends on using simple or structured value
... Do we want to be able to give first-name, last-name, etc...

Hui: strong type is better

we can define a URI type, constraining the value to be a valid URI

Chris: should getCreator return a URI, or a structured value with different properties ?

Pchampin: I think that if we decide to provie a structured value, it should be reusing existing vocabularies.

<raphael> But your URI will point to a structured object ?

<raphael> ... foaf:Person?

<raphael> ... ebu:Person?

Chris: returning a URI shifts the problem to the application.

<raphael> ... iptc:Person ?

<raphael> many choices :-)

<veroniqueM> what about vocabularies that have no URI?

<chris> well, it is not clear for the moment so the URI allows all of them

<veroniqueM> lists of person names in thesauri

<raphael> ... DIG35 also defines Person :-)

<fsasaki> I second veronique's question

<raphael> veroniqueM, what's their type ?

<raphael> we are interested in the type rather than the instance

<veroniqueM> skos:COncept at best

<veroniqueM> but usually instances of

<raphael> and how would you do to get more information about the person?

<veroniqueM> instances of skos:Concept

<veroniqueM> not skos:Concept themselves

<chris> +q

Pchampin: really to independant questions?
... 1/ do we want to embed description or point to them (URI)?
... 2/ do we prescrive a given vocabulary?

<veroniqueM> I think that with the mapping table, we list a set of possible vocabularies

<veroniqueM> but we should not limit the value to these

<raphael> I'm not sure I understand what you meant veroniqueM: http://dbpedia.org/resource/Barack_Obama rdf:type skos:Concept means ... Barack Obama is a skos:Concept

Chris: I think we should define a minimal vocabulary for persons.

<veroniqueM> in that case yes

Raphael: let us ask Jean-Pierre his own rationale for defining new types or resusing existing ones

PChampin: any one to write that mail to Jean-Pierre? :)

<Hui> Here's a module from BONDI : http://bondi.omtp.org/1.01/apis/contact.html ; it tries to abstract the AddressBook in various mobile phones. They created detailed structure about contact. Anything we can learn from their approach?

Raphael: Chris could also elaborate on this. Jean-Pierre could tell us about that choice in EBU.

<tobias> I can send this email

<tobias> ...because this is a very interesting question, actually.

ACTION Tobias to ask Jean-Pierre about the rationale in EBU to redefine the notion of Person rather than reuse an existing one

<trackbot> Created ACTION-156 - Ask Jean-Pierre about the rationale in EBU to redefine the notion of Person rather than reuse an existing one [on Tobias B├╝rger - due 2009-09-29].

Chris: a couple of existing vocabularies would fulfill our needs.

(dig35, mpeg-7 probably)

But it would not be a good idea to have out API returning a structured value from another ontology.

<chris> dig35

<veroniqueM> Raphael just explained me what the discussion was about :0

<veroniqueM> and I would vote for creating a ma:person

<veroniqueM> linked to foaf:person

<veroniqueM> to define things in our namespace

<veroniqueM> like we did for the equivalents of dc:title etc

<tobias> yes I agree

PChampin: if we use a generic structure (property-value paires), it is acceptable to populate it with properties from other ontologies

Chris: in favour of defining a concept

<chris> how will we define the "things"?

Pchampin: me too

<chris> do you mean with "things" specific person-related concepts like first_name, last_name, ...

<raphael> My approach would be to start from the use cases, model what is necessary, and no more

Pchampin: but do you intend to do the same for *properties* of those "secondary" concepts?

<Hui> things probably just DOMString

<raphael> don't try to model everything you could know about a Person if it is useless

<chris> I agree

<raphael> Again, Jean Pierre has a very precised idea of what a metadata is necessary to define a Person in broadcasted TV content use case

<raphael> ... we could start from that


<veroniqueM> well, the idea is to model what is present in the Multimedia standards, so another option is to strat from there, the same as we selected the core properties of the Media Ontology, no?

<tobias> I agree to both of Raphael's comments.

<raphael> who wants to compare all xxx:Person ?

<chris> I proposed more or less the same as what your typing now Veronique

<raphael> [adjourned]

<tobias> @Raphael: I have read a comparison of the definitions of person in existing standards some years ago. perhaps I find it somewhere and will send it arounf

<raphael> it would be useful Toby, thanks!


now there here

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2009/09/22 13:03:55 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.135  of Date: 2009/03/02 03:52:20  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/by Felix/by Felix (at a workshop)/
Succeeded: s/in the coming week/by the end of the week/
Succeeded: s/Florian: we now/Chris: we now/
Succeeded: s/xx45?/dig35/
Succeeded: s/out namesapce/our namespace/
Succeeded: s/stanards/standards/
Found Scribe: Pierre-Antoine Champin
Found ScribeNick: pchampin
Found ScribeNick: pchampin
Default Present: +, tmichel, pchampin, tobias, Hui, florian, Felix, raphael, Chris, wonsuk
Present: + tmichel pchampin tobias Hui florian Felix raphael Chris wonsuk
Agenda: http://lists.w3.org/Archives/Public/public-media-annotation/2009Sep/0041.html
Found Date: 22 Sep 2009
Guessing minutes URL: http://www.w3.org/2009/09/22-mediaann-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]