W3C

- DRAFT -

MAWG F2F day 2

05 Nov 2010

See also: IRC log

Attendees

Present
jean-pierre, yannick, daniel, joakim, florian, tobias, wonsuk, werner, thierry, philippe_le_hegaret
Regrets
Chair
daniel
Scribe
wbailer, wbailer_, Joakim

Contents


<wbailer_> scribenick: wbailer

implementation and test suite

<wbailer_> florian presents implementation at http://mawg.joanneum.at

<wbailer_> image gallery demo and more technical demo available

<wbailer_> scribenick: wbailer_

florain: user selects image, loads metadata resources, then possible to query the properties

florian: next step: integrate ontology
... add trust layer to handle collission between metadata (e.g. dc creation date vs exif creation date)
... other next steps could be to use it to connect to youtube or flickr api and map to ma
... question is how to build test suite around implementation
... which kind of test cases should we have

werner: two levels of tests: correct return types, then correct mapping

yannick: query properties from different formats
... do we need to support all formats?

conclusion at last meeting was: two implementations overlapping with some formats, but not for all

daniel: someone has to create metadata for tests

<stegmai> http://mawg.joanneum.at/ <- reference implementation

florian: link is on sourceforge

daniel: how to add additional metadata format?

florian: you need to implement extractor and mapping for a format

joakim: did you write your own extractor?

florian: in this case it's only SAX parser and XPath

joakim: we do not want to have the metadata extractors part of the tests
... should we have an intermediate format for metadata being extracted

werner: only for non-xml format? what is the added value, we have to avoid that there is already mapping in the intermediate format

florian: test suite should test api spec, thus not the mapping, but the functionality of the api
... do we need one test suite for ontology + api, or two, one for mapping and one for the api
... these are two different steps, that can be tested separately
... problem with intermediate format: how to store the metadata information

wonsuk: we need to define sample test files for the different formats
... depending on the implementation we can make a program that processes the files

florian: problem, who creates further implementations
... do not plan to add further formats myself

[wonsuk draws proposed workflow for tests on flip chart]

wonsuk's proposal: have a matrix of test cases and formats, specify methods to be called and arguments, and expectedd response (e.g as JSON)

thierry: do we need to have test for all of them?

philippe: would expect that one of them is complete

jean-pierre: source of complete implementation should be published

thierry: we have to define the criteria to exit cr

philippe: the way test was done for infoset was to say that other specs use it
... you have to prove that mapping works
... no need to have software for ontology doc
... implementations of api are needed
... question is how many of the properties need to be supported in the api
... for ontology, propose to have one file for each format, with the same metadata in it, plus target mappings to ontology
... publish source and target files

joakim: files cannot be identical, because different formats have different scopes

philippe: publish service, that eg uses EXIF extractor and creates RDF according to defined ontology

[florian draws api test for web service use case on flip chart]

florian: separating api and ontology tests would help people who just want to use the ontology

yannick: api and ontology tests can be linked when the rdf files that are results of ontology tests are used as api tests

philippe: propose to decouple timelines of ontology and api doc, move ontology doc forward as fast as possible
... to enforce getting the mapping examples, put formats at risk to push people

joakim: start from list of people who initially did the mappings to assign responsibilities for providing mappings

original mapping responsibilities http://www.w3.org/2008/WebVideo/Annotations/wiki/Format_mapping_review

thierry: this is not the final list, it's missing the container formats

frbr has already been removed

table in the wiki has been updated

<scribe> ACTION: tobias to provide example RDF files for the mapping target [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action01]

<trackbot> Created ACTION-355 - Provide example RDF files for the mapping target [on Tobias Bürger - due 2010-11-12].

<tmichel> zakim this is Team_(mediaann)

[coffee break]

joakim: after tobias has provided example, each responsable for a format to provide a complete mapping file

tobias will provide template within 2 weeks

deadline for data: jan. 10, 2011

then have a round of reviews of the proposed files (similar to as we did for the mapping tables)

should be finished before F2F

thierry: to go cr: (1) handle comments in 2nd LC, (2) disposition of comments document, (3) state cr exit criteria and discussion with director

(3) could be scheduled in Jan

we could leave cr for ontology doc after Feb. F2F

joakim summarises next step for test suite, will put text on wiki and send it by mail

[lunch break]

<danielp> API discussion issues: Wonsuk's proposal, Robin's comments, Doug's comments, etc...

<danielp> More precise of Rating description.

<danielp> action to Wonsuk to make more precise description of rating in ontology

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

<danielp> action to wonsuk to make more precise description of rating in ontology

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

<danielp> ACTION: Wonsuk to make more precise description of rating in ontology [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action02]

<trackbot> Created ACTION-356 - Make more precise description of rating in ontology [on WonSuk Lee - due 2010-11-12].

<danielp> next is Wonsuk's proposal http://lists.w3.org/Archives/Public/public-media-annotation/2010Nov/0059.html

<Joakim> The description of the rating in the ontology doc must be more precis

API design proposal from Wonsuk

<Joakim> W want to simplify the number of calls, to get the values for a properties

<Joakim> Minimize the number of parameters

<Joakim> and merge getPropertyNamesHavingValues() and getMediaProperty

<Joakim> Werner agrees to 1) and 3) in the email,

<Joakim> If you pass an empty array you get all elements

<Joakim> Werner: regarding 2) not sure. Maybe easier to keep it as it is

<Joakim> following the comment from Dough, making the interface cleaner, and not having a long list

<Joakim> Wsk: To use an object parameter, IS easier (Wonsuk agrees)

<Joakim> Ask browser implmenters point of view regarding 2)

<Joakim> Wnsk: In Opera have object type of parameters with optional filters

<Joakim> the code is more self explanatory with solution in 2)

<Joakim> We wait for agreements from chris and the rest of the group

<Joakim> Florian agrees with 3) and 1)

<Joakim> Werer; how would this affect sub-types?

<Joakim> Wonsuk: we can use "dot-notation"

<Joakim> When you call getProperty for more than one properties, what to do with sub-properties

<Joakim> Ignore all, or return sub-types of the same cardinality

<Joakim> This is about the "filter"

<Joakim> Resolution: The WG decides wether to implment these minor changes after WG consencus via email

<scribe> scribenick: Joakim

<scribe> ACTION: chris restructure the list of "interfaces", as discussed on Thursday 4/11 [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action03]

<trackbot> Created ACTION-357 - Restructure the list of "interfaces", as discussed on Thursday 4/11 [on Chris Poppe - due 2010-11-12].

<scribe> ACTION: chris Add descriptive information to some sections for example 3.1.1, and what is an "interface" [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action04]

<trackbot> Created ACTION-358 - Add descriptive information to some sections for example 3.1.1, and what is an "interface" [on Chris Poppe - due 2010-11-12].

<scribe> ACTION: chris Add type and value in all return types [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action05]

<trackbot> Created ACTION-359 - Add type and value in all return types [on Chris Poppe - due 2010-11-12].

action Florian To provide an example of an automatic test routine (before x-mas break)

<trackbot> Created ACTION-360 - Provide an example of an automatic test routine (before x-mas break) [on Florian Stegmaier - due 2010-11-12].

action Joakim To ask web browser vendors for feed-back on API spec

<trackbot> Created ACTION-361 - Ask web browser vendors for feed-back on API spec [on Joakim Söderberg - due 2010-11-12].

<danielp> ACTION: joakim to ask browser vendors for their opinion on the mawg documents. [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action06]

<trackbot> Created ACTION-362 - Ask browser vendors for their opinion on the mawg documents. [on Joakim Söderberg - due 2010-11-12].

action chris change the type from short to double value, min and max

<trackbot> Created ACTION-363 - Change the type from short to double value, min and max [on Chris Poppe - due 2010-11-12].

<scribe> ACTION: chris changed the description in 3.14.2 target audience interface; the name should be "identifier", and rating scheme should be used [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action07]

<trackbot> Created ACTION-364 - Changed the description in 3.14.2 target audience interface; the name should be "identifier", and rating scheme should be used [on Chris Poppe - due 2010-11-12].

Chris draft respons to Robin

How to introduce in HTML5?

Werner: Can we really say that user agents must expose MediaResource interface

we should drop the example rather

wonsuk: in the video element, they provide properties concerning the media that are not relevant in our spec

he suggest to drop HTML5 example, and replace it with the existing ones

<scribe> ACTION: joakim respond to the email in the archive 2419, from Paul Cotton [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action08]

<trackbot> Created ACTION-365 - Respond to the email in the archive 2419, from Paul Cotton [on Joakim Söderberg - due 2010-11-12].

<danielp> In order to help Joakim: contact Frank Olivier (microsoft), Chris Double (Mozilla), Eric Carlson (Apple), Philip (Opera) + Macjek + Paul Cotton + Sam Ruby and cc Mike Smith and Philippe Le Hegaret

The only way we can have this strong statement (2419-a) we need the group integrate our interface

<danielp> just cut-and-paste by raphael's last email

respons to 2419-b

sync/async API

The sync API is motivated by our Web-service use case

werner: Suggest splitting the interface into two: sync and async

wonsuk: Add section in the specification: Sync Interface (for web service implmenters) and Async Interface (for web developers)

<scribe> ACTION: chris (with help of API editors) Add section in the specification: Sync Interface (for web service implmenters) and Async Interface (for web developers [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action09]

<trackbot> Created ACTION-366 - (with help of API editors) Add section in the specification: Sync Interface (for web service implmenters) and Async Interface (for web developers [on Chris Poppe - due 2010-11-12].

respons to 2419-b -2

Having filters?

suggest to keep filters!

respons to 2419-c

This ambiguity will be solved be the proposed changes in the API. If not implemented Chris reply is ok

respons to 2419-d

the change has been accepted!

respons to 2419-e

The WG agrees with suggested respons

respons to 2419-f

WG suggest to formulate a new response that reflects the proposed API changes from wonsuk

response to 2419-g

<danielp> ACTION: spark3 to verify what's the most exact word as (CreateDate or CreationDate) to a native [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action10]

<trackbot> Created ACTION-367 - Verify what's the most exact word as (CreateDate or CreationDate) to a native [on Soohong Daniel Park - due 2010-11-12].

response to 2419-h

we need to be more clear what we mean with compression in the ontology doc

To clarify what information is stored in ma:format vs ma:compression, and give example for a jpg file or raw mpeg2 file

action joakim improve description text for ma:compression and ma:format

<trackbot> Created ACTION-368 - Improve description text for ma:compression and ma:format [on Joakim Söderberg - due 2010-11-12].

response to 2419-i

WG agrees with respone

response to 2419-j

WG agrees with respone

response to 2419-k

<wonsuk> Ilkka Oksanen, Nokia

Wonsuk spoke to editor of capture API (Ilkka, DAP), asking them to use our API in order to handle metadata

<danielp> http://www.w3.org/TR/capture-api/

action joakim In 2nd LC, send spec to Ilkka. One UC: creation-date, creator technical properties, location

<trackbot> Created ACTION-369 - In 2nd LC, send spec to Ilkka. One UC: creation-date, creator technical properties, location [on Joakim Söderberg - due 2010-11-12].

The group work 30 min on AP

CME presentation

<wbailer_> see slides sent by daniel

<wbailer_> CME is a membership org, specs specified and available to members

<wbailer_> they still need to define how to work with other organisations such as W3C

<wbailer_> CME is semantic web technology, is basically a serialised RDF graph

<wbailer_> use existing ontologies

<wbailer> authorization/authentication mechnisms (e.g. use proof of purchase of electronic release to access premium content)

<wbailer> designed to be multi-platform

<wbailer> CME influences by MusicOntology and FRBR

<wbailer_> joakim: do you use ID3?

<wbailer_> gregg: we use UITS, a signed code, that is proof of purchase signed by the seller, included in ID3

<wbailer_> ... identifier from ID3 might be useful

<wbailer> use media library to synchronise between devices and access media from different formats

<wbailer> seeking members, opportunity for invited experts

<wbailer> gregg: difficult to share actual spec, but the most important properties are in the slides

<wbailer> ... could work out with steering committee how to work with mawg

<wbailer> joakim: are you finished with your architecture?

<wbailer> gregg: current v0.9 spec, work in progress, ontology are the most mature parts

<wbailer> joakim: which mawg could be integrated in CME?

<wbailer> gregg: you tried to normalise metadata used in different format, in that sense CME is yet another container, but some concepts (e.g. technical are similar)

<wbailer> ... reduced amount of metadata to reduce burden on authors

<wbailer> gregg: on divergence between CME and MAWG is modeling of contributor/creator

<wbailer> werner: do not see so many differences

<wbailer> gregg: you tried to do a very flat ontology

<wbailer> werner: we do not have expression, we do not model the frbr distinctions

<wbailer> gregg: interested in media fragments

<wbailer> pa: scope of MF is to give URIs to fragments

<wbailer> joakim: do you expect your spec to be adopted

<wbailer> gregg: available through click through license

<wbailer> ... authoring of packages should be made easier

<wbailer> gregg: metadata is one form of premium content, but rather enables other forms of premium content, such as better recommendations

<wbailer_> gregg: metadata is added value, but rather enables further premium content such as better recommendations

<wbailer_> ... hope that authors add better metadata

<wbailer_> ... premium content can be earlier release of concent videoes

<wbailer_> daniel: question to CME: requirements from CME?

<wbailer_> gregg: work more towards better practices to annotate

<wbailer_> ... integrate media ontology wiht music ontology

<wbailer_> gregg: more info on cme is on website

next F2F

<wbailer_> scheduled in week of Feb. 14

<wbailer_> contacted Apple for hosting

<wbailer_> alternative is VU Amsterdam

<wbailer_> to be decided at upcoming telco

<wbailer_> [adjourned]

<danielp> next teleconf is November 16th. (No teleconf in next week)

Summary of Action Items

[NEW] ACTION: chris Add descriptive information to some sections for example 3.1.1, and what is an "interface" [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action04]
[NEW] ACTION: chris Add type and value in all return types [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action05]
[NEW] ACTION: chris (with help of API editors) Add section in the specification: Sync Interface (for web service implmenters) and Async Interface (for web developers [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action09]
[NEW] ACTION: chris changed the description in 3.14.2 target audience interface; the name should be "identifier", and rating scheme should be used [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action07]
[NEW] ACTION: chris restructure the list of "interfaces", as discussed on Thursday 4/11 [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action03]
[NEW] ACTION: joakim respond to the email in the archive 2419, from Paul Cotton [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action08]
[NEW] ACTION: joakim to ask browser vendors for their opinion on the mawg documents. [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action06]
[NEW] ACTION: spark3 to verify what's the most exact word as (CreateDate or CreationDate) to a native [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action10]
[NEW] ACTION: tobias to provide example RDF files for the mapping target [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action01]
[NEW] ACTION: Wonsuk to make more precise description of rating in ontology [recorded in http://www.w3.org/2010/11/05-mediaann-minutes.html#action02]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2010/11/05 15:53:20 $

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/emial/email/
Succeeded: s/wokr/work/
Found ScribeNick: wbailer
WARNING: No scribe lines found matching ScribeNick pattern: <wbailer> ...
Found ScribeNick: wbailer_
Found ScribeNick: Joakim
Inferring Scribes: wbailer, wbailer_, Joakim
Scribes: wbailer, wbailer_, Joakim
ScribeNicks: wbailer, wbailer_, Joakim
Present: jean-pierre yannick daniel joakim florian tobias wonsuk werner thierry philippe_le_hegaret
Got date from IRC log name: 05 Nov 2010
Guessing minutes URL: http://www.w3.org/2010/11/05-mediaann-minutes.html
People with action items: add chris descriptive information joakim respond spark3 tobias wonsuk

[End of scribe.perl diagnostic output]