RE: [Minutes] GGIE (Glass-to-glass internet ecosystem) Task Force meeting, 22nd April 2015

Just some thoughts about content tagging:
•       There should be a differentiation between Content and Assets that are related to that Content. Content is an abstract concept for an entity. That entity can be consumed as content via various media asset forms.
•       Content should be identified with a globally unique ID (e.g. URN) that is extensible
o       Items related to a Content ID should have identifiers that can build on the Content ID to create an extended globally unique identifier that points back to the root Content by ID. Such items could include metadata, assets, and sub-content extracted from the original content.
o       Relationships of items to a Content ID could also be indicated in metadata, but that is a less immediate method for rich identity and would not support authentication of the content as efficiently.
o       Identifiers for Sub-content (e.g. clip extracts from a video) might include the source content ID and an offset indicator field to describe where in the source content it was extracted from. This could be applied to temporal media (audio/video,) images and text, and more.
o       Authorized use of content would include fully relational rich identifiers for authentication or a pointer to an authentication certificate. Unauthorized use would not have valid IDs or certificates.
o       Authentication is important not only for authorization but for legal acceptability in a world where digital content is an integral part of the resulting augmented reality of a digital society.
•       Short cuts like embedding metadata in ID tags are useful in a closed context (e.g. local content management) but make global consistency and semantic alignment more difficult.

Thanks!

-Paul


-----Original Message-----
From: Andrew Zamler-Carhart (azamlerc) [mailto:azamlerc@cisco.com]
Sent: Wednesday, April 22, 2015 1:57 PM
To: public-web-and-tv@w3.org
Subject: Re: [Minutes] GGIE (Glass-to-glass internet ecosystem) Task Force meeting, 22nd April 2015

Hi folks,

Here are a few questions that came to mind during the call.

Basic premise: video content should be uniquely tagged at the time that it is created.

Should the tag be associated with the user/operator, the camera/device, the studio, and/or the production?

How would these entities (user, device, studio, production) be identified?

Does a tag need to be unique to the user, device, or studio? Or does it need to be globally unique? (I would say the latter.)

Who can get the information about each video? Can anyone lookup a tag to find the name of the creator? Or would it be the responsibility of the user to assert their ownership of tagged content? Should the user have control over this?

Does the device need to make a request to a central authority in order to make a tag? Or can it autonomously create a UUID that is unique in time and space, even when the device does not have internet access, and perhaps register it later?

What are the important differences between consumers who post home videos to YouTube and Hollywood movie studios? Are we targeting both? Do we treat them the same? Prosumer productions are blurring the boundaries, with amateurs using pro hardware and professionals using iPhones.

What is the relationship between clips that are tagged during the recording process and finished videos that are prepared for distribution?

Photos are discrete but video is continuous. Would several clips made in quick succession be considered the same content? If a clip is cut in half, are the two clips considered different?

Multiple videos can be combined in many ways as part of the editing process. Will derivative works contain all the information from component clips? Example 1: I download two videos from YouTube and combine them with a cross-fade to make a new video. The creators of the first two videos should be able to assert ownership of their content. Example 2: My studio makes a movie composed of thousands of takes. Viewers should probably not be able to identify the name of the cameraman who shot each take.

How resilient to tampering or accidental removal does the tagging need to be? Example 1: If the tag is stored as part of the file’s metadata, it can be useful for those who want to identify the video. Can it survive transcoding by various apps? Example 2: If the creator wants to tag a video in such a way that the tag can not be easily removed, can it survive intentional efforts to remove it?

Resolving some of these questions (such as watermarking algorithms) may be out of scope for GGIE, but we may still identify a need for some of these issues to be solved by other parties.

Cheers,
Andrew

> On Apr 22, 2015, at 12:55, Daniel Davis <ddavis@w3.org<mailto:ddavis@w3.org>> wrote:
>
> Hello all,
>
> Here are the minutes for today's GGIE Task Force conference call:
> http://www.w3.org/2015/04/22-webtv-minutes.html

>
> and pasted in full below.
> Thanks to Bill Rose for scribing again.
>
> Also, please note that the next call will be on Wednesday 20th May
> (NOT 6th May due to the W3C AC meeting).
>
> With regards,
> Daniel Davis
> W3C
>
> ==========
>
>                Web and TV Interest Group Teleconference
>
> 22 Apr 2015
>
>   See also: [2]IRC log
>
>      [2] http://www.w3.org/2015/04/22-webtv-irc

>
> Attendees
>
>   Present
>          glennd, Bill_Rose, paul_higgs, azamlerc, ldaigle,
>          digitaldale, NiloMitra, MarkVickers, ddavis
>
>   Regrets
>          Kaz
>
>   Chair
>          Glenn_Deen
>
>   Scribe
>          Bill_Rose
>
> Contents
>
>     * [3]Topics
>     * [4]Summary of Action Items
>     __________________________________________________________
>
>   <ddavis> trackbot, start meeting
>
>   <glennd> thank you
>
>   <trackbot> Date: 22 April 2015
>
>   <ddavis> Grrrr
>
>   Glenn called the meeting to order at 11:04
>
>   Glenn did a role call.
>
>   <azamlerc> Hi everyone! Andrew Zamler-Carhart at Cisco.
>
>   Bill Rose taking notes
>
>   <ldaigle> maybe did not leave last time?
>
>   <glennd> [5]http://www.w3.org/2015/04/08-webtv-minutes.html

>
>      [5] http://www.w3.org/2015/04/08-webtv-minutes.html

>
>   <ddavis> scribenick: Bill_Rose
>
>   Glenn posted link to the notes. Minutes were approved without
>   change.
>
>   <glennd> Use Review:
>   [6]https://www.w3.org/2011/webtv/wiki/GGIE_TF/UseCases/Content_

>   Capture#Use_Cases
>
>      [6]
> https://www.w3.org/2011/webtv/wiki/GGIE_TF/UseCases/Content_Capture#Us

> e_Cases
>
>   Glenn led a discussion on Content Capture Use Cases (see link)
>
>   Question: Discussing professional, non-prof or both?
>
>   Glenn: Lines are blurred. There are still differences but hard
>   to draw a line between.
>
>   Dale: IDs are generated for downstream consumption and there
>   are differences in consumption between pro and amature.
>
>   Glenn: Looking at YouTube there is no hard line in consumption
>   and workflows to post and consume.
>
>   Dale: The identifiers are produced for a usage. The association
>   of that data may be heavier in pro compared to consumer.
>
>   Glenn: Proposed approach - lump together until/unless a
>   difference is identified.
>
>   Dale: Good approach.
>
>   Glenn: Review Content Capture UCs
>   ... UC-1 Basic Video Asset Capture
>
>   Assumes fixed bit-rate and resolution; stored on device; single
>   file or container with both audio and video with metadata about
>   video, camera, capture event, user, etc.
>
>   Glen: Step 5 in description is a mistake and should not be
>   included.
>
>   Glenn: Typically ID is added (e.g. YouTube) at time of
>   publication. Similar to adding SKU # by the store before
>   shelving.
>   ... Some services can capture live video, upload to service for
>   posting. ID is added by service.
>
>   Correction: Some "users" can capture live video +"e.g. on a
>   phone"...
>
>   Glenn: This happens in near real time.
>
>   Dale: Can see this happening in ad insertion situation.
>
>   Glenn: Can also happen in social environment - e.g. a sports
>   event, etc., where each stream uploaded by individuals are all
>   associated with the same event but get different IDs added due
>   to different user captures.
>   ... Capture-UC-2 Assigning a unique content identifier at asset
>   capture
>   ... User captures a/v scene. Device obtains a content ID from
>   an issuing service.
>
>   Dale: Might the ID add an ID (if it does not have connection to
>   an issuing service) and later get a unique ID from the service?
>
>   Glenn: A UUID could be issued to the camera by an issuing
>   authority that is used by the camera to generate a unique
>   content ID for each video it records.
>
>   Dale: There could be multiple delegations - YouTube, Facebook,
>   EIDR, etc. Device could be pre-registered for each service it
>   posts to.
>   ... Could this be extended to assemblages of content -
>   associated text, etc., which becomes part of the "content".
>
>   Glenn: That might be an extension, more sophisticated use case
>   ... Think of this as a network of associated content where
>   multiple streams can be assembled into a composite stream for
>   final consumption
>
>   Dale: All of the various streams are associated through the IDs
>   for potential use in a composite - e.g. tributary streams that
>   can be assembled into final composited stream - potentially at
>   the end under direction of some actor - client, etc.
>
>   Glenn: New actor for the extended discussion above - the
>   composer, sponsor, user, etc., that does the composing function
>   to collect and bundle the tributary streams for consumption.
>
>   <scribe> ACTION: Item to Create new UC for the composite stream
>   concept discussed above. Include new actor. [recorded in
>   [7]http://www.w3.org/2015/04/22-webtv-minutes.html#action01]
>
>   <trackbot> Error finding 'Item'. You can review and register
>   nicknames at <[8]http://www.w3.org/2011/webtv/track/users>.
>
>      [8] http://www.w3.org/2011/webtv/track/users

>
>   Andrew: Do we need to describe how the compositing process
>   would work including the Identifiers that allow different
>   composites to be uniquely identified
>
>   Glenn: Think it is necessary to identify the composites
>   uniquely.
>
>   Andrew: 2 ways to look at it: Things useful to editor(s) and
>   those that are useful to the user.
>
>   Glenn: gets into rights management which is out of scope.
>
>   Bill Rose: User could subscribe to a particular social user's
>   composite.
>
>   Glenn: Could be done by using a prefix associated with that
>   social user.
>
>   <scribe> ACTION: Item to Dale, Andrew, Bill - write UC to
>   describe the scenarios we discussed [recorded in
>   [9]http://www.w3.org/2015/04/22-webtv-minutes.html#action02]
>
>   <trackbot> Error finding 'Item'. You can review and register
>   nicknames at <[10]http://www.w3.org/2011/webtv/track/users>.
>
>     [10] http://www.w3.org/2011/webtv/track/users

>
>   Leslie: Do we need to have a concept of an identifier of a
>   user/composer versus the identifier of a composite. Seem to be
>   mixing the two.
>
>   Glenn: Can have multiple prefixes for multiple '
>
>   personas'
>
>   <ldaigle> agree re. privacy
>
>   Glenn: Need to look at privacy here as well. How do we describe
>   this to include privacy. Instinct is there will be 2 concepts -
>   globally unique prefixes (permanent) and locally unique
>   prefixes that can be 'thrown away"
>
>   WebTV GGIE
>
>   Glenn: Think of it as the difference between a cell phone that
>   is bound by phone # to the user and a throw away phone that is
>   only loosely bound to the user not the users permanent phone #.
>
>   Andrew - Que: How do you distinguish between several "takes" of
>   the same scene versus different events on the same camera?
>
>   Glenn: Opportunity to explore how this could be handled but it
>   is out of scope for GGIE.
>
>   <scribe> ACTION: Item to Andrew - write up the question (above)
>   for posting. [recorded in
>   [11]http://www.w3.org/2015/04/22-webtv-minutes.html#action03]
>
>   <trackbot> Error finding 'Item'. You can review and register
>   nicknames at <[12]http://www.w3.org/2011/webtv/track/users>.
>
>     [12] http://www.w3.org/2011/webtv/track/users

>
>   Glenn: Capture-UC-3 Fingerprinting and associating a content
>   identifier at asset capture
>
>   <ldaigle> @Bill — are you unmuted?
>
>   I am muted
>
>   <azamlerc> Folks, I need to run... I'll write up some of the
>   questions that I've raised. Cheers!
>
>   <ldaigle> who just sneezed....
>
>   <digitaldale> I will work to document the use case
>
>   <ldaigle> and sniffed
>
>   Glenn: UC-3 expands on UC-2 by generating a digital fingerprint
>   and registering it along with the unique content ID
>   ... Content ID is obtained from the issuing authority which is
>   accitated with the fingerprint. Not a simple hash.
>   ... Can generate same video fingerprint even with different
>   audio or video encodings
>   ... Why not a watermark? For final products watermarks work
>   well. But not appropriate for capture because watermarks can
>   alter the original capture that could degrade the capture.
>   Fingerprint does not.
>
>   <digitaldale> watermarking may be better introduced downstream
>   which is tied to a commecial registration authority such as
>   EIDR or AD-ID
>
>   Glenn: Downstream services can still add watermark for many
>   reasons but for capture case a fingerprint might be more
>   useful/acceptable.
>   ... Gap identied in each UC is there is no issuing authority
>   for the Content Identifier, prefixes, etc. that can be
>   referenced for these purposes.
>
>   <digitaldale> this way there would not need to be watermark
>   extraction or rewatermarking if the watermark is introduced at
>   the capture device and then reinserted later downstream
>
>   Glenn: In UC-3, no standardized fingerprint algorithm for this
>   purpose, and no issuing authority that can register the
>   fingerprint with the content ID
>   ... Any other business?
>
>   Daniel: Want to encourage more participation in this work. If
>   anyone has any ideas to improve/broaden participation let us
>   know. Would like to do a monthly mailing on the activities
>
>   Que: Can we alert User Groups (Community Groups) to this work?
>
>   Daniel: Will look into how this can be done.
>
>   Perhaps LinkedIn Groups?
>
>   <ddavis> [13]https://www.w3.org/2011/webtv/wiki/GGIE_TF

>
>     [13] https://www.w3.org/2011/webtv/wiki/GGIE_TF

>
>   <ddavis> [14]https://www.w3.org/2011/webtv/wiki/Main_Page

>
>     [14] https://www.w3.org/2011/webtv/wiki/Main_Page

>
>   <ddavis>
>   [15]https://www.w3.org/2011/webtv/wiki/GGIE_TF#Participation

>
>     [15] https://www.w3.org/2011/webtv/wiki/GGIE_TF#Participation

>
>   Glenn: Next meeting is scheduled for May 6th which coincides
>   with W3C meetings in Paris. Should we skip that meeting or
>   reschedule?
>   ... Propose skip May 6th and schedule May 20th as next meeting.
>
>   paul_higgs: If not addressing rights management, how about
>   policy relating to RM?
>
>   <ldaigle> LInked Content Coalition
>
>   Glenn: If we describe how rich metadata could be associated
>   with content, that could be used to point/link to RM services,
>   etc.
>
>   Daniel: On administration, we will be shifting to WebEx
>   sometime in June so the bridge will be changing around June.
>
>   <ldaigle> thanks! bye.
>
>   Glenn: Next meeting May 20th. Closed the meeting at 12:24 ET.
>
>   <ddavis> Meeting: Web and TV Interest Group Teleconference:
>   GGIE TF
>
>   <ddavis> Meeting: Web and TV Interest Group Teleconference -
>   GGIE TF
>
> Summary of Action Items
>
>   [NEW] ACTION: Item to Andrew - write up the question (above)
>   for posting. [recorded in
>   [16]http://www.w3.org/2015/04/22-webtv-minutes.html#action03]
>   [NEW] ACTION: Item to Create new UC for the composite stream
>   concept discussed above. Include new actor. [recorded in
>   [17]http://www.w3.org/2015/04/22-webtv-minutes.html#action01]
>   [NEW] ACTION: Item to Dale, Andrew, Bill - write UC to describe
>   the scenarios we discussed [recorded in
>   [18]http://www.w3.org/2015/04/22-webtv-minutes.html#action02]
>
>   [End of minutes]
>     __________________________________________________________
>