W3C home > Mailing lists > Public > public-media-annotation@w3.org > February 2009

Media Annotations WG teleconf minutes 2009-02-03 (DRAFT)

From: Felix Sasaki <fsasaki@w3.org>
Date: Tue, 03 Feb 2009 23:27:37 +0900
Message-ID: <49885459.9040502@w3.org>
To: "public-media-annotation@w3.org" <public-media-annotation@w3.org>

... are at http://www.w3.org/2009/02/03-mediaann-minutes.html and below 
as text.



[1] http://www.w3.org/


Media Annotations Working Group Teleconference

03 Feb 2009



See also: [3]IRC log

[3] http://www.w3.org/2009/02/03-mediaann-irc


Davy, Felix, Veronique, Joakim, Tobias, Raphael, Thierry,
pchampin, wonsuk Werner

frank, jean_pierre, victor, michele




* [4]Topics
1. [5]action items
* [6]Summary of Action Items

<joakim> hello

<joakim> [7]http://www.w3.org/2009/01/27-mediaann-minutes.html

[7] http://www.w3.org/2009/01/27-mediaann-minutes.html

<joakim> scribe?

<scribe> scribe: fsasaki



<scribe> chair: joakim

<wbailer> i can scribe next week

action items


[9] http://www.w3.org/2008/WebVideo/Annotations/track/actions/open

AI 43 - pending

AI 47 - pending

felix: the document
amicmedia_partnerguide.pdf resolves this AI I think


<raphael> Is it what you are talking about:
.pdf ?

[11] http://www.adobe.com/devnet/xmp/pdfs/DynamicMediaXMPPartnerGuide.pdf


<raphael> my uri is shorter :-)


close ACTION-49

<trackbot> ACTION-49 Check with Adobe about what properties of XMP
are actually used closed

<scribe> ACTION: Joakim with Felix to read
.pdf and use it to see how XMP is actually used [recorded in

[12] http://www.adobe.com/devnet/xmp/pdfs/DynamicMediaXMPPartnerGuide.pdf

<trackbot> Created ACTION-80 - With Felix to read
.pdf and use it to see how XMP is actually used [on Joakim Söderberg
- due 2009-02-10].

[14] http://www.adobe.com/devnet/xmp/pdfs/DynamicMediaXMPPartnerGuide.pdf

AI 52 open, jean-pierre not here

AI 54, AI 55 , kept open

close ACTION-66

<trackbot> ACTION-66 Do searchmonkeymedia closed

AI 71 - ongoing, Thierry sent a mail , no response yet

close ACTION-71

<trackbot> ACTION-71 Ask AC rep of IBBT to nominate Chris closed

close ACTION-77

<trackbot> ACTION-77 Merge the contributions to the mapping table on
the wiki closed

close ACTION-78

<trackbot> ACTION-78 Start a discussion on the wiki about the DC vs.
DCterms issue closed

<VeroniqueM> dc:title skos:exactMatch vra:title

<raphael> alternate: dc:title owl:equivalentProperty vra:title

raphael: if such a statement would be formally written in a document
we need to say what it means in terms of semantics
... we need to be careful about whether the example matches the
exact semantics of the properties

felix: how does this example relate to requirement 11?

<raphael> req 11:

[15] http://www.w3.org/TR/2009/WD-media-annot-reqs-20090119/#req-r11

veronique: an RDF-way for representing relationsships

felix: so this might be one slice of requirement11?

<raphael> sameAs is for instances

veronique: yes

<raphael> equivalentClass is for classes

<raphael> equivalentProperty is for properties

raphael: current phrasing of r11 says that prose is a MUST, RDF
serialization is a MAY
... we might get a comment that the RDF serialization is a MUST. My
comment was if we use an RDF serialization we should make clear what
the consequences are

<raphael> +1 for having an rdf serialization (sw bias)

veronique: group should vote whether we should have that

<tobias> +1 (sw bias, too)

<raphael> You could also let the comments from the FPWD coming

<raphael> Felix: the question is not about rdf serialization or not,
but rather a must or may

<VeroniqueM> my question would be: what do you do for the
vocabularies that do not have a standard rdf serialisation yet?

<raphael> ... an API who reads only metadata does not necessarily
need to embed an RDF parser

<raphael> Veronique: what are the standards in the current mapping
table that do not have an RDF serialization ?

<wbailer> a formal representation would allow to specify to express
relations and constraints more clearly, the API has to conform to
the semantics defined by the ontology - by whatever means the API
implements that

<pchampin> +1 wbailer

<raphael> +1 for werner

<pchampin> and the underlying RDF does not have to influence the API

<pchampin> (meaning: return values, parameters)

<pchampin> I might as well

<tobias> +1 for werner

felix: who is planning to implement the api?

<joakim> I will probably

<wbailer> we will probably

probaly raphael with tobias, joakim, werner, felix

<VeroniqueM> how many of the vocabularies from the mapping table
have an RDF version?

felix: who will make RDF processing a mandatory part of the API


<tobias> Veronique: I guess approx. 10

joakim: some people want to use RDF, but they don't want to make it

<VeroniqueM> ok, thanks!

<raphael> Veronique: DC, XMP, ID3, MediaRDF, EXIF, DIG35, FRBR, VRA,
IPTC, EBU, MPEG-7, DMS-1 at least

<pchampin> I have a problem with the notion of "RDF serialization"

wonsuk also wants to implement API

<pchampin> what is interesting about RDF, IMHO, is the conceptual

<wbailer> +1 pchampin, it's actually not about serialization

<pchampin> the clear semantics of RDF vocabularies is the
interesting features

<VeroniqueM> +1

<raphael> @PA: indeed, there is no rdf serialization, there is an
XML syntax, a turtle syntax, a n3 syntax, etc.

felix: who wants to handle in an implementation problems like
semantic mismatch (e.g. different serializations of a date) and
different contents (e.g. author as a string vs URI)

raphael: we will have these problems anyway at some point
... it has to be in the spec or it has to be dealt with in the

felix: should we deal with this or should the problem be left to the

raphael: if you do that in a standardized way implementation cost is

felix: do you mean our standard or a different one?

raphael: both .. we need to be able to say "values are of type XYZ"
in our spec

<wbailer> + 1, it needs to be in the specification (either one type
or set of few options if we go for a more complex api)

joakim: seems not so much effort to put this into a standard

<wonsuk> +1

<tobias> +1 for raphael

felix: who would implement the basic reading of metadata, e.g.
reading XMP metadata?

joakim: I thought for XMP reading we would rely on an organization
who is responsible for the format?

werner: for many formats there are APIs

<raphael> +1 for werner

<wbailer> would it make sense to start collecting information on
available apis for the different formats and their capabilities?

<raphael> Felix: my concern is really about what would be the test
cases that would make the CR exit criteria for the spec

felix: I thought that we would use the API for testing, but if we
have no basic reading mechanism for formats, how to implement the

<raphael> I think it is a good idea to start listing all tools and
api that can process specific formats,

<raphael> ... proprietary or not

<pchampin> +1 raphael

+1 to raphael

<pchampin> I think there are 2 kinds of things that the test cases
could/should check

<joakim> +1

<pchampin> 1st thing: conformance with the API itself

<VeroniqueM> +1 Raphael

<pchampin> 2nd thing: consistency with the underlying metadata

<wonsuk> +1 for raphael

<raphael> @pchampin: you would need to define 'conformance' and

<joakim> Felix: implement some few formats deep rather than many

felix: I want to try to work deeply with a few properties, e.g.
provide maybe even reading of the related properties, rather than
having a lot of properties in our ontology
... but not providing the basic means for reading

joakim: have the feeling that people want to use RDF for
implementing the toy application

<pchampin> not necessarily (for me)

for me yes as one toy application, but not the only one

<pchampin> rather to specify the mappings more formally, hence the

<pchampin> but the implementation does not have to rely on RDF
libraries, in the end

veronique: there is quite a number of existing formats which is in
RDF already

<joakim> Felix: the test cases are independent of the implementation
(RDF, XML etc.)

joakim: we should focus on defining the test cases, independent of
specific implementations anyway

<VeroniqueM> to Felix: Raphael answered DC, XMP, ID3, MediaRDF,
EXIF, DIG35, FRBR, VRA, IPTC, EBU, MPEG-7, DMS-1 at least

<pchampin> I can try

joakim: pa had some idea, if he could implement that we can see if
that helps our needs

<pchampin> me too :-D

<pchampin> ok

<pchampin> it will depend on the libraries that are available

<scribe> ACTION: Pierre-Antoine to make a toy implementation of
mappings [recorded in

<trackbot> Created ACTION-81 - Make a toy implementation of mappings
[on Pierre-Antoine Champin - due 2009-02-10].

<VeroniqueM> vra has already a mapping to DC in RDF, as subclasses

raphael: we have asked media fragments WG to make a review
... we provided something, felix provided an answer, do you expect
more reviews?

<tobias> I guess Michaels comments are very valuable already

felix: would just be good to have some more mail exchange with
Michael or others

raphael: plan to also do the review, probably in mid February

<raphael> bye

<joakim> bye

joakim: thanks to all, adjourned

<wonsuk> +1 for Fbye

Summary of Action Items

[NEW] ACTION: Joakim with Felix to read
.pdf and use it to see how XMP is actually used [recorded in
[NEW] ACTION: Pierre-Antoine to make a toy implementation of
mappings [recorded in

[17] http://www.adobe.com/devnet/xmp/pdfs/DynamicMediaXMPPartnerGuide.pdf

[End of minutes]

Minutes formatted by David Booth's [20]scribe.perl version 1.133
([21]CVS log)
$Date: 2009/02/03 14:27:11 $

[20] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
[21] http://dev.w3.org/cvsweb/2002/scribe/
Received on Tuesday, 3 February 2009 14:35:06 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 February 2009 14:35:07 GMT