Media Annotations Working Group Teleconference -- 08 Sep 2010

<trackbot> Date: 08 September 2010

<scribe> Meeting: Media Annotations Working Group Face to Face Meeting

Opening

Daniel introducing the face to face meeting goals

<scribe> scribe: raphael

<scribe> scribenick: raphael

Thierry: I have put all comments received on the mailing list in the tracker
... goal is to resolve these comments
... I will then send an answer to all commenters

Thierry explaining the disposition of comments process

<wonsuk> http://www.w3.org/2006/02/lc-comments-tracker/42786/WD-mediaont-10-20100608/

<wonsuk> http://www.w3.org/2006/02/lc-comments-tracker/42786/WD-mediaont-api-1.0-20100608/

Thierry asking if public observers can attend

Joakim: yes, no problem

Thierry: request from Alexandre Monin to attend the f2f meeting, works in IRI (Pompidou), about the timelines project: http://www.iri.centrepompidou.fr/outils/lignes-de-temps-2/
... there are more information in French

<tmichel> MA WG charter

<tmichel> http://www.w3.org/2008/01/media-annotations-wg.html

All the room is discussing various comments received

<tmichel> Raphaêl: should give the exact status of this Ontology

s/Rapha�l/Raphael

<vero> ... and explicit what is cast in stone: a minima the name of the properties

LC comments for the Media Ontology

9 comments received: http://www.w3.org/2006/02/lc-comments-tracker/42786/WD-mediaont-10-20100608/

<tmichel> Timed Text Markup Language (TTML) 1.0

<tmichel> http://www.w3.org/TR/2010/CR-ttaf1-dfxp-20100223/

<tmichel> Guidelines for Using W3C Last Call Tracker Tool.

<tmichel> http://www.w3.org/2008/WebVideo/Annotations/guidelines-tracker-tool.html

LC Comment 2389

URI: http://www.w3.org/2006/02/lc-comments-tracker/42786/WD-mediaont-10-20100608/2389

We can have external subtitles

scribe: ma:relation seems to be the right way for doing this

the relation would be 'subtitle'

scribe: the URI would be a srt file or a WebSRT file or a TimedText file

Looking at whether ma:fragment could be use to embed a subtitle within a description

WG would not recommend to do this

if ma:fragment is to be used, then the URI is a W3C Media Fragment URI

scribe: and the role is the actual free text description

Raphael: I suggest to rename the 'role' attribute of ma:fragment into 'description'

<tmichel> 3 steps: Subtitles are in a different file and Ma: rematiuon is the right way

<tmichel> or embeded in the media like a track

<tmichel> and third is to use ma:fragment

s/Ma: Rematiuon/ma:relation

Raphael: for the second case, ma:numTracks should be used

Thierry: when do we draft answers? I would like the group approve the answers before they are out

Joakim: there is a hacking session tomorrow

<scribe> ACTION: werner to draft an answer for the comment 2389 [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action01]

<trackbot> Created ACTION-270 - Draft an answer for the comment 2389 [on Werner Bailer - due 2010-09-15].

Raphael: for semantic annotation, we cannot have yet a RDF graph as a value for a media fragment
... can we?

Werner: the problem is to have this inline
... if your triples live in a separate file, then use ma:relation
... then relation would be 'annotation'

and the URI would be the location of the RDF file

scribe: we don't have yet a way to embed this annotation in the media description

<tmichel> Raphael: we should also look into Web SRT

Werner: ma:description is useful for free text description
... should we allow complex semantic description with this property?

Wonsuk: I think ma:relation is more appropriate

<tmichel> Raphael: could use another property like ma:semantic description

<tmichel> Raphael: should coordinate and ask expertise from with the Semantic Coord Group for this kind of property

<tmichel> impact will be to review all the mapping tables

<scribe> ACTION: troncy to contact the SW Coordination Group about how to do semantic annotation with ma:description and ma:relation alternatives [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action02]

<trackbot> Created ACTION-271 - Contact the SW Coordination Group about how to do semantic annotation with ma:description and ma:relation alternatives [on Raphaël Troncy - due 2010-09-15].

Summary: first part will be addressed by Werner, while the second part will require input from SWCG via Raphael

LC Comment 2403

http://www.w3.org/2006/02/lc-comments-tracker/42786/WD-mediaont-10-20100608/2403

<tmichel> Raphael: we could have a new informative section to show how to use the ontology in other W3C technologies like POWDER, RDFa, etc.

Raphael: very good idea from Phil
... we could take his example as it is
... I would even add a ma:description property in the descriptorset with the value being the same than the <*displaytext>

Joakim: we need to have a new section with Examples
... such as in the API document, see http://www.w3.org/TR/2010/WD-mediaont-api-1.0-20100608/#usage-examples
... we need to have a new section with Relationships with other standards such as POWDER

<scribe> ACTION: wonsuk to start a new section in the Ontology document providing example of usage (e.g. subtitles, powder, etc.) [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action03]

<trackbot> Created ACTION-272 - Start a new section in the Ontology document providing example of usage (e.g. subtitles, powder, etc.) [on WonSuk Lee - due 2010-09-15].

LC Comment 2393

http://www.w3.org/2006/02/lc-comments-tracker/42786/WD-mediaont-10-20100608/

Raphael: the question is that several container formats such as WebM, MPEG4, etc. define properties such as frameRate, description, etc.
... the question is: should they be included in the mapping table?
... which media container formats to include?
... are they spec available?

WebM spec: http://www.webmproject.org/code/specs/container/

Joakim: the list of metadata formats covered by the mapping table is not closed

Werner: perhaps we could do something separate for all container formats
... WebM, MPEG4, etc.
... we don't want to have an exhaustive list of technical properties

Joakim: it makes sense to have a different section for these formats ... since we will map only few properties as we are not interested in technical properties, see Use Case and Requirements document

Raphael: Speex seems to be a ogg container
... I suggest to have a list of media container formats concerned by this new section

<tmichel> Raphael: if you cover ogg you cover Speex

Raphael: most likely: OGG, MPEG4, WebM, flv, avi?, 3gp?

Veronique: for OGG, Silvia has made a mapping

<ChrisP> MPEG21?

Werner: mxf?
... mov

Chris: do we do something for the other media? still images?

Werner: we have already EXIF ... we could argue that there are some jpeg with no exif and relevant metadata
... but I suggest to not go this way

Joakim: we need to have a volunteer to make this new section with mapping tables for the media container format

<tmichel> raphael: for media container format mappings, Keep just title and the technical properties

Joakim: mapping table will include the technical properties and title|description
... who volunteers for an action?

<ChrisP> some refs for container formats: http://en.wikipedia.org/wiki/Comparison_of_container_formats

Raphael: another ref from the Media Fragments WG: http://www.w3.org/2008/WebVideo/Fragments/wiki/State_of_the_Art/Containers

<scribe> ACTION: Thierry to prepare the mapping table for the media container formats [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action04]

<trackbot> Created ACTION-273 - Prepare the mapping table for the media container formats [on Thierry Michel - due 2010-09-15].

Raphael: sometimes there is an empty row, sometimes there is a N/A
... what is the semantics of N/A

what is the semantics of the empty row?

Room agrees to put N/A in the second column as for the OGG mapping table

<scribe> ACTION: thierry to check all mapping tables and add N/A when it is needed [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action05]

<trackbot> Created ACTION-274 - Check all mapping tables and add N/A when it is needed [on Thierry Michel - due 2010-09-15].

<tmichel> Put N/A in the second column

<tmichel> ... in all the mapping tables.

<tmichel> Comment LC-2398

<tmichel> Veronique to draft a response about our light weith approach.

<tmichel> Thierry will update the response.

scribenick vero

<vero> ACTION: Joakim and Wonsuk to deal with with the editorial comments of the LC comment 2405 [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action06]

<trackbot> Created ACTION-275 - And Wonsuk to deal with with the editorial comments of the LC comment 2405 [on Joakim Söderberg - due 2010-09-15].

<vero> Raphael: for the ontology implementation/description, what should be done is

<vero> ... make clear prose to define classes, properties of the ontologies, the values that they can have

<vero> ... be sure that there are no different classes/properties with same intended semantics

<vero> ... provide the implementation that we are currently doing as THE RDF/OWL version of the ontology, leave it open for an modeling in other languages

<vero> ... in the new version of the document, add the promised appendix (the ontology's appendix) and link your comments to the appendix

<vero> Joakim: check out the status of the appendix when we go through the doument, because this section has to be updated

<vero> ... based on Raphael's comment

<vero> Joakim: we have to add in the Media Ontology document that the list of properties/classes and mappings described are normative, and so is the RDF/OWL version of the ontology that is provided in the appendix

<vero> ... in future work

<vero> ... we have to specify in the section that points to the appendix that the RDF/OWL is the authoritative representation of the Media Ontology

<vero> Doug's LC comments point to the "column syntax", column is used as a prefix for the namespace and also in the Type definition, but for another semantic

<vero> Raphael's proposal: keep ma: as a prefix and change the syntax of the type definition to attribute name/attribute value pairs in prose

<vero> ACTION: Joakim to change the type definition into the new syntax [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action07]

<trackbot> Created ACTION-276 - Change the type definition into the new syntax [on Joakim Söderberg - due 2010-09-15].

<vero> ACTION: vero to check out whether we do not say anymore that "ma" is the namespace, but that it is the prefix for it, in the Ontology document [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action08]

<trackbot> Sorry, couldn't find user - vero

<vero> ACTION: Veronique to check out whether we do not say anymore that "ma" is the namespace, but that it is the prefix for it, in the Ontology document [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action09]

<trackbot> Sorry, couldn't find user - Veronique

<daniel> NEW AGENDA ITEM: RDF TF will be discussed from 3 - 4 PM today.

<vero> LC comment: the ontology applies only to stored digital medias, and the document should specify it

<vero> Werner: this is not totally accurate, as long as the resource can be identified by a URI, it is OK

<vero> ... can be a tape, can be offline, it does not matter

<vero> LC comment: PBcore should be added to the mapping list and the original broadcast date should be added in the property list

<vero> Werner: PBcore could be added

<vero> Werner: second point can be dealt with with ma:createDate

<vero> Chris: agree could also provide an example

<vero> Werner: shall we invite her to do the mapping and submit it?

<vero> Joakim: +1

<vero> Thierry: +1

<wonsuk> +1

<vero> ACTION: Chris to add a line for the "broadcastDate" property attribute [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action10]

<trackbot> Created ACTION-277 - Add a line for the "broadcastDate" property attribute [on Chris Poppe - due 2010-09-15].

<vero> LC comment 2411 3rd bullet: more complex content description and time-dependent description

<vero> Werner: same answer as for the "embedded subtitle" case for 2nd part, and we have to decide on a way of representing semantic descriptions... can also point her to the fact that there are separate properties for keywords, genre too

<vero> LC comment 2411 follow-up: include also preservation metadata?

<vero> Joakim: out of scope

<tmichel> Tobias, The Zakim bridge numbers are

<tmichel> US: +1.617.761.6200 ,

<tmichel> FR: +33.4.26.46.79.03

<tmichel> UK: +44.203.318.0479

<tmichel> with code 26632 (CONF2)

<tmichel> I guess it is better to use the US number. We are connected to it ...

Tobias is online, so we're discussing the status of the RDF implementation of the Media Ontology

<vero> Tobias: agrees to have one authoritative RDF version

<vero> ... we should align the version I am working on with Jean-Pierre with Raphael's implementation

<vero> ... we did not implement the mappings yet, but the vocabulary modeling is quite stable now

<vero> ... implemented PIerre-Antoine's comments, except one that is still open discussion

<vero> ... in my opinion, it can be published in W3C space

<vero> Werner: what is the rationale for the subclassing of creator and contrinbutor

<vero> Werner: what is the rational for the Actor class, because it's not part of the Media Ontology document, we should be careful if this represents the authoritative RDF version of the ontology

<vero> Tobias: subclasses represented the roles that the actors or organizations can play

<vero> Werner: problem is that in EBUcore, the classes are disjoint if I recall correctly, and in MPEG7 the sublassing is dont in the inverse way

<vero> ... so we should discuss this

<vero> Tobias: indeed Actor is not in the ontology

<vero> Werner: so we have to precise what is normative and what is not

<vero> Daniel: can you give us an update on the model?

<vero> Tobias: we fixed domain errors, discussed on the list the fact that some properties should not apply to certain type of media document (like: doe snot apply to images)

<vero> ... Pierre-Antoine commented on that

<vero> ... proposed to define an image as a media document with a duration 0

<vero> ... which can be something else, though

<vero> ... Jean-PIerre proposed the notion of "isMadeOf"

<vero> ... latest version of the ontology is at http://www.salzburgresearch.at/~tbuerger/ma-ont-rev7.owl

<vero> ... mapping is not dealt with yet, the problem with the mappings is that you need more than OWL to model them, you need rules

<vero> Werner: we modeled some of these rules

<vero> ... datatypes mappings are dealt with in a second step

<vero> ... and structural changes are hard to express

<vero> Vero: it owuld be useful for the clarification of the Media Ontology document to have two instances of media documents and implement the properties like ma:related

<vero> to give the reader concrete examples of how this is represented, particularly for the "complex types"

<vero> Tobias: would be useful, but do you have an idea of what language to use for these annotations?

<vero> ... for the mappings, we can refer to services that do the mappings

<vero> Werner: we check for which property in the source file maps to which property in the target format/file, and then we do datatype conversions etc

<vero> The mappings do not have to be implemented in the RDF representation of the Media Ontology, all of the properties yes

<vero> Joakim: we have to check out that we do not include in the RDF ontology properties/classes that are not in the Media Ontology document

<vero> Chris: and the other way around: we have to make sure that all the Ontology Document properties and classes are covered in the RDF implementation

<vero> Joakim: how should we format the appendix document?

<vero> Thierry: if I get the OWL file I can include it in the Media Ontology document

<vero> Thierry: when can we have a version that is "includable"?

<vero> Tobias: next week is ok

<vero> Thierry: great, then we can indeed link the comment's answers to the RDF ontology in the appendix

<tmichel> the RDF Implementation OWL ma-ont-rev7 from the RDFS Taskforce is now available at

<tmichel> http://www.w3.org/2008/WebVideo/Annotations/drafts/ontology10/CR/ma-ont-rev7.owl

Raphael: Hum, I'm trying to follow some of the modeling decision, there is one thing I disagree: an image is NOT a video of duration 0
... we discussed this in the Media Fragments WG, a video of duration 0 is a video
... not a still image
... don't change the mime type
... HTML5 has the same rationale
... and they invent the poster for extracting an image of a video

<tobiasb> hr raphael, "an image is a video of duration 0" is not modeled in our current version as Vero said, JP and myself disagreed to the proposal from PA

I'm happy yo agree with Tobias and Jean Pierre on this :-)

scribe: it is not that usual

ReDeFer is made by Roberto Garcia and Oscar Celma, maintained by the former

scribe: but I'm talking about the script that Yves, Dan, etc. have made
... to generate MusicOntology, FOAF, SIOC, etc.

I have the perl script on my machine, not easy to throw on irc :-)

scribe: but I'm sure google will tell you where it is ;-)

Find it!

Name = SpecGen

Original version from @dajobe, part of redland, http://crschmidt.net/semweb/redland/

<scribe> New version hosted on DERI: http://sw.deri.org/svn/sw/2005/08/sioc/ontology/spec/specgen4.py

UNKNOWN_SPEAKER: there is python and prolog version

<ChrisP> scribenick ChrisP

http://www.w3.org/2006/02/lc-comments-tracker/42786/WD-mediaont-10-20100608/2417

<ChrisP> comment 2417

<ChrisP> on copyright and policy properties

<ChrisP> David Singer has initiated some comments on this on the mailing list

<ChrisP> tmichel: we keep copyright as is

<ChrisP> veronique: we will use the proposed refinements for the type

<vero> then the existing: { (policy:URI|String), (type:String)? }

<vero> becomes { (identifier:URI), (statement:String), (type:String)? }

<vero> as attributes for the ma:policy property

<ChrisP> werner: we should respond that we understand that there is an overlap

<ChrisP> werner: but there is a need from the industry to keep the seperate copyright

<vero> and we need it for the copyright property mapping

<ChrisP> ACTION: Werner to draft a response to this comment [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action11]

<trackbot> Created ACTION-278 - Draft a response to this comment [on Werner Bailer - due 2010-09-15].

<ChrisP> joakim: we should integrate the given examples in the document

<ChrisP> werner: we an ask to provide more information on the given examples

http://www.w3.org/2006/02/lc-comments-tracker/42786/WD-mediaont-10-20100608/2418

<ChrisP> comment 2418

<ChrisP> first part are editorial comments

<ChrisP> there is a problem with the CSS rule

<ChrisP> ACTION: Thierry to look into problems with the CSS rule [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action12]

<trackbot> Created ACTION-279 - Look into problems with the CSS rule [on Thierry Michel - due 2010-09-15].

<ChrisP> veronique will tackle some of the editorial comments

<ChrisP> joakim: we have to check the normative and informative parts of the document

<ChrisP> ACTION: Joakim to check the normative parts of the document according to comment 2418 [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action13]

<trackbot> Created ACTION-280 - Check the normative parts of the document according to comment 2418 [on Joakim Söderberg - due 2010-09-15].

<ChrisP> a substantive comment is made by robin about a statement in the document stating that the definitions in the ontology document MAY change

<ChrisP> compatibility with the API document should not be a problem

<ChrisP> what about the claimed compatibility with the HTML 5 W3C Working Draft

<ChrisP> werner: Florian has mentioned something related to the name collisions with HTML 5

<ChrisP> ACTION: Werner to follow up the comment of Robin about the HTML 5 compatibility issue by contacting Florian [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action14]

<trackbot> Created ACTION-281 - Follow up the comment of Robin about the HTML 5 compatibility issue by contacting Florian [on Werner Bailer - due 2010-09-15].

<ChrisP> ACTION: Wonsuk to follow up the comment on the namespaces [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action15]

<trackbot> Created ACTION-282 - Follow up the comment on the namespaces [on WonSuk Lee - due 2010-09-15].

<ChrisP> to deal with the second substantial comment we will change SHOULD to MUST

<ChrisP> to deal with the third substantial comment we will change SHOULD to RECOMMENDED

<ChrisP> next issue is on the description of the location property

<ChrisP> is there a controlled vocabulary for a coordinate system

<ChrisP> werner: we could give an example of a good use of the location property

<vero> example of geolocation vocabulary

<vero> http://www.geoconcept.com/WGS84.html

<vero> http://www.georss.org/Main_Page

<ChrisP> werner: maybe ask Geo Location WG for an example, or a good controlled vocabulary

<ChrisP> ACTION: Daniel to send mail to Geolocation WG for an example, or controlled vocabularies that can be used for the ma:location property [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action16]

<trackbot> Sorry, couldn't find user - Daniel

<wbailer> http://www.iana.org/assignments/media-types-parameters

<ChrisP> next question is whether ma:format includes media type parameters

<ChrisP> it is not really clear what is meant here

<ChrisP> ACTION: daniel to send mail to Geolocation WG for an example, or controlled vocabularies that can be used for the ma:location property [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action17]

<trackbot> Sorry, couldn't find user - daniel

<daniel> ACTION: daniel to send mail to Geolocation WG for an example, or controlled vocabularies that can be used for the ma:location property [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action18]

<trackbot> Sorry, couldn't find user - daniel

<daniel> ACTION: spark3 to send mail to Geolocation WG for an example, or controlled voca that can be used for the ma:location property [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action19]

<trackbot> Created ACTION-283 - Send mail to Geolocation WG for an example, or controlled voca that can be used for the ma:location property [on Soohong Daniel Park - due 2010-09-15].

<ChrisP> werner: the ma:format allows the MIME type as a string so media type parameters could be added

<ChrisP> joakim: this is a topic for David

<ChrisP> ACTION: werner to mail David Singer about the inclusion of media type parameters in the ma:format property [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action20]

<trackbot> Created ACTION-284 - Mail David Singer about the inclusion of media type parameters in the ma:format property [on Werner Bailer - due 2010-09-15].

<ChrisP> ACTION: joakim to check the normative statemens according to the editorial comment of Robin [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action21]

<trackbot> Created ACTION-285 - Check the normative statemens according to the editorial comment of Robin [on Joakim Söderberg - due 2010-09-15].

<ChrisP> we will fill the XPATH column in for every table

<ChrisP> ACTION: thierry to update the XPATH columns of the mapping tables [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action22]

<trackbot> Created ACTION-286 - Update the XPATH columns of the mapping tables [on Thierry Michel - due 2010-09-15].

<ChrisP> some information should be added to the introduction of 4.2.2 to describe the tables and columns

<ChrisP> ACTION: Joakim to include information to the introduction of 4.2.2 to describe the tables and columns [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action23]

<trackbot> Created ACTION-287 - Include information to the introduction of 4.2.2 to describe the tables and columns [on Joakim Söderberg - due 2010-09-15].

<ChrisP> ACTION: felix to unify the XPATH expressions in the mapping tables according to the last substantial comment of Robin [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action24]

<trackbot> Created ACTION-288 - Unify the XPATH expressions in the mapping tables according to the last substantial comment of Robin [on Felix Sasaki - due 2010-09-15].

Guys, for a good location vocabulary, there is the W3C one :-) WGS84 !!! http://www.w3.org/2003/01/geo/

<ChrisP> ACTION: Joakim to deal with the last editorial comment of Robin [recorded in http://www.w3.org/2010/09/08-mediaann-minutes.html#action25]

<trackbot> Created ACTION-289 - Deal with the last editorial comment of Robin [on Joakim Söderberg - due 2010-09-15].

<ChrisP> only one comment is left

<ChrisP> will be scheduled first thing tomorrow

<ChrisP> the LC Comments on the API document are also postponed to tomorrow

Have a good dinner guys if you're adjourning for today

- DRAFT -

Media Annotations Working Group Teleconference

08 Sep 2010

Attendees

Contents