[annodex-dev] Timed URI seeking without mod_annodex

[Ryan Shaw]

illiminable wrote:

> I also really like the idea of "playing the cmml" because to me... 
> that's the more html-like way... annodex is the equivalent or packaging 
> up all the html, js, css and images in one single bigass file. And it 
> also allows what i think will be a great use of cmml and that is to 
> annotate media you either don't own or don't want to host. ie... some 
> video on your site... i can write a cmml file and put it on my site... 
> and the player will get the cmml from me and the media from you. Which 
> the server side solution can only support by double handling the media.
> 
> It also lets multiple people and places annotate the same single remote 
> media file. ie some video somewhere... 10 different people can do their 
> own annotation of the same single copy of the media. Not only does this 
> allow more freedom... think about it from a search engine perspective... 
> with the current solution you get one persons annotations of the 
> media... if you seperate the two... you can have dozens of people 
> annotating the same file, with their own opinions/persectives/commentary.
> 
> That's the sort of thing that will get people in at "entry-level"... 
> since it's trivial for me to host a cmml file... and not have to worry 
> about the bandwidth of serving the video, or copyright issues... just 
> like html... i can write up my spiel and link it to the original source, 
> rather than everyone duplicating the source media everywhere they want 
> to annotate it.

My research team at UC Berkeley is currently applying for funding to 
pursue a project much like what you describe, which we have dubbed Media 
Streams Metadata Exchange. We have been considering Annodex as the 
platform on which we would build, but one of our design constraints is 
that it must be possible to annotate audiovisual documents openly and 
distributedly--there can be no remote server requirement for annotation 
other than read-access to the document.

Since our research mainly focuses on the design of metadata schemes for 
describing media and socio-technical systems for evolving these schemes, 
rather than the nuts and bolts of how descriptions and media files are 
linked, we would like very much to build on the great work you have 
done. But the current architecture, which puts control of whether and 
how media is annotated in the hands of the authors or publishers, is not 
suitable for what we have in mind.

We have discussed an architecture like what Andre suggested, which uses 
seek tables to enable client-side seeking. Our idea was that seek tables 
could be hosted completely independently of the original media content. 
The first MSMDX-enabled player to encounter a new media file would 
create a seek table and upload it to the MSMDX service, where it would 
be indexed by a hash (maybe a Bitzi Bitprint) of the file. Later other 
MSMDX-enabled players could retrieve the seek table via this hash.

Anyway, I understand that you may have different goals for the Annodex 
project, but since the ideas Andre and Zen have posted seem so close to 
what we have envisioned, I thought I should introduce myself so we might 
discuss this further.

Cheers,

Ryan Shaw
Garage Cinema Research
http://garage.sims.berkeley.edu/