25 Feb 2015

See also: IRC log


JP, PaulG-ATT, alex_deacon, MikeW, mark_vickers, azamlerc, kaz, glennD, ldaigle, NiloM, BillRose


<glennD> Agenda for todays call

<glennD> 1. Approval of meeting minutes from 2/11/2015

<glennD> 2. agenda bash

<NiloM> +1.913.841 is NiloM

<glennD> 3. Use Cases in Wiki review & discuss

<glennD> 4. next set of cases to focus on

<glennD> https://www.w3.org/2011/webtv/wiki/GGIE_TF/UseCases#Use_Cases

Glenn did a role call by region

<azamlerc> Hi folks. Andrew Zamler-Carhat, Cisco France

<ldaigle> glenn

<ldaigle> your audio is busted

Glenn had to hang up and call back in to fix audio

<azamlerc> Andrew Zamler-Carhart, Cisco France

Bill Rose is acting as scribe

<ldaigle> Leslie Daigle is here on the east coast

Glenn called for the approval of last meeting minutes - there were approved without change

Use Cases discussion

<JP> link again?

Glenn called for everyone to pull up the link to the Use Cases for discussion. It includes a template.

<azamlerc> http://www.w3.org/2011/webtv/wiki/GGIE_TF/UseCases

Glenn asked for everyone to review and comment on the Contents, terminology, etc.

Glenn will assign the identifiers and keep them unique and orderly

Glenn populated the WiKi with 3 Use Cases to start the discussion

Glenn discussed the first Use Case - UC-1 Basic Steamed Content Viewing

Glenn - trivial UC. It does not discuss how it is done, buffering, etc. Not real world. Mainly identifies the actors as an exercise in writing a UC.

Glenn - UC-2 Streamed Content Viewing. This is a bit more realistic.

Glenn - instead of a single URL a manifest is returned by the source to the app listing multiple sources

Glenn - app (sender) is in control instead of the receiver. App can control flow rate, buffers, fill rate, bit rates, etc. for optimization

Nilo - Is this to describe adaptive streaming?

Glenn - yes, one of the things introduced is adaptive streaming. Are there changes needed to the UC?

Nilo - can be read in several ways - same content is stored in multiple locations yet it appears to imply the files may have different representations.

Glenn - thinking was that so much work has been done on ABR (e.g. DADH) that we don't need to go into those details. More for exposing actors and concepts. Would it be helpful to introduce more?

Correction - DASH, not DADH

Glenn - will add more on what could be in the manifest to clarify

Glenn - UC-3 Identifying Streamed Content

<JP_Intel> Sorry, can someone indicate where these UCs can be retrieved?

Glenn - builds on UC-2 but adds viewer interactions. Viewer selects an asset from the list sent by sender and send request for specific content

<NiloM> http://www.w3.org/2011/webtv/wiki/GGIE_TF/UseCases

Glenn - key point is to obtain an identifier

<JP_Intel> Thank you

Glenn - identifier describes what is being played but not format, resolution, etc. Different versions will have different identifiers e.g. EIDR but no other information about the file.

Glenn - If 2 identifiers are equal, then the content is equal in terms of sequence of frames, frames included, etc. but no info on format properties.

Question - what might the differences be between 2 "equal" IDs?

Glenn - 2 copies of a Directors cut of a movie may be equal but in different resolutions - SD/HD. Based on how EIDR works today

Glenn - different compression could be used.

Nilo - In example of 2 Directors Cut versions, are you saying SD/HD may have same EIDR?

Glenn - It may have same ID but does not have to. Depends on what is registered with EIDR. Arranging to have someone from EIDR to talk to this group and answer questions. Posted link to EIDR for the group.

Andrew - Useful to look at EIDR as a starting point but not the only one to look at. Is is a flat or hierarchical? Does not seem to include enough info.

Glenn - their view is not hierachical in that it does not include e.g. geo info. Issue a new EIDR for different versions. But they maintain additional info for each EIDR registered that can be accessed.

Glenn - long ID has more information such as registrant name but much of the info is not included in the ID.

Glenn - described 3 ways to extract unique ID - from metadata; from imbedded watermark (audio or video); or from fingerprint

Glenn - 2 things to do with extracted ID - record/report for measurement/logging; Form a URI to be used with other video content sources to uniquely identify the content being viewed

Glenn - URI can be used to locate additional caches of the content to improve viewing experience

Glenn - today there is no univeral standard way to ID content, use fingerprint to ID, create URI based on an ID, etc.

<mark_vickers> EIDR example (click on "Relationships" tab to see hierarchy): https://ui.eidr.org/view/content?id=10.5240/6933-25C9-299D-671A-24FB-V

Glenn - Action Item; Create new UC where extraction is the first step

Action Item- ? List metadata included in the watermark by creator at creation and added downstream by the work flow

<trackbot> Error finding 'Item-'. You can review and register nicknames at <http://www.w3.org/2011/webtv/track/users>.

Glenn - We will need to capture how to identify whether a cache has the entire file versus portions of the file at e.g. a CDN? Is anyone willing to take this UC as an action item/

Question - From user perspective I search and get back a set of URIs for the content which can be used to obtain the content, switch to a new URL if there is an interuption, etc.

Glenn - Yes this is one of the goals of GGIE. The metadata can be used to resolve information associated with content for use throughout the workflow, user, delivery, etc. Current CDNs have a basket of "tricks" to aid in the delivery of the content. But they have very little info from the client.

Glenn - part of the idea here is to discuss information that the client could contribute to improve CDN performance.

<JP_Intel> Sorry, Glenn, but I need to drop now due to a conflcit. Will discuss offline by e-mail about introducing what you need for Content Identifier extraction.

Leslie - Glenn is describing how best to find the optimum version for delivery to the client/user. But I think I heard Nilo ask how does a user find the "Directors Cut" itself?

Nilo - seems there are several levels of resolution required - "I want a copy of the Directors Cut"; Second level of resolution includes optimizing locating the optimal flow from cache to client.

Glenn - so user experience is search/discovery - not network layer. What is available to watch?

? - Also what is available from your service(s)/provider(s) versus what is available in general.

Glenn - yes. Could also include other formats, languages, etc., which may be aviable on a different service on a different device.

Andrew - - Yes - same content might also be available from multiple providers I have access to. Would like to see all and choose myself.

Glenn - Anyone willing to write a rough flow of several of these concepts. Nilo, Andrew?

Nilo, Andrew - yes.

Nilo - already available using e.g. Google.

Nilo - I thought this is what GGIE is focused on.

Andrew - more technical people may be capable of using Google but many are not.

Glenn - The stuff from Google contains what they have discovered and put into results they think you want to see. Different from accurate/complete results tailored to the user's actual services, clients, etc. GGIE can contribute to this area of a user centric focus.

Glenn - about all content including user content, not just commercial content

Andrew - not describing world today, but what it could be

Glenn will try to populate additional UCs. Anyone can contact him to add to the UCs.

Glenn - next meeting: March 11, 11:00 AM ET.

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.140 (CVS log)
$Date: 2015-02-25 23:37:44 $