W3C SW Coordination Group Telco

07 Apr 2010

See also: IRC log


Tom, Michael, Sandro, Chris, Manu, Scott, Ivan, Raphaƫl, Joakim


<manu-work> http://connectedmediaexperience.org/

Introductions: MichaelH, ManuS, Raphael, Tom, Scott, ChrisW, Joakim...

Joakim: senior associate at Ericsson, Sweden, works with multimedia indexing - co-chair media annotation WG
...Introductions: Sandro (and of course Ivan)


<ivan> minutes of last meeting

Resolved: minutes from 2010-03-10 accepted.

Next meeting: 2010-04-21

Media Annotation WG update

Joakim: Have been working for 1.5 years, now close to Last Call for our ontology and API spec.
... More or less successfully managed to map 20 formats and core set of ?20 attributes
... Want to make sure we have not missed any. Want to specify "exact", "close", "related".
... We also consider datatypes. Not trivial.

<manu-work> URL to latest editors draft?

Joakim: We have mapping table. Recently query from BBC re: RDF implementation of ontology.
... Many are keen on SW technologies, so Task Group in group to pursue this as parallel activity - not in charter.

<raphael> Ontology for Media Resources 1.0: http://www.w3.org/TR/2010/WD-mediaont-10-20100309

<raphael> API for Media Resources 1.0: http://www.w3.org/TR/2010/WD-mediaont-api-1.0-20100309

Joakim: Working on API defined with Web ADL

<raphael> Use Cases and Requirements document: http://www.w3.org/TR/2010/WD-media-annot-reqs-20100121

Joakim: put out current specs - have review from Opera and others.
... would like to have review process finished in order to hold Last Call in April or May

<Zakim> manu-work, you wanted to ask about Connected Media Experience.

Manu: Does the MA prefix in section 4.1.2. map to a known URL, described by RDF document - do we expect people would be able to use this ontology directly?

Ivan: Use ma:creator or dc:creator?

Joakim: Spec currently in prose, so no RDF implementation or defining a namespace. Currently not on the agenda. Will not be part of the recommendation.

Ivan: Doesn't tell me how to add metadata in RDF.

Tom: Jisc undertook a vocabulary mappiung framework, how does it relate?
... i am unusure about what mapping initiatives try to achieve
... do you know about that?

joakim: I did not know about that initiative
... reference?

<mscottm> http://www.jisc.ac.uk/

TomB: my general question is: from a sw point of view we have a landscape of vocabularies, and now we have these mapping initiatives
... and the question is what the function of those mappings is supposed to be

<manu-work> Vocabulary mixing is also going to become very prevalent when RDFa 1.1 hits the web next year.

TomB: is it something that augments the use of distributed vocabularies, or does it replace them?

joakim: we did discuss that
... some members would like to see a specification with rdf
... it is still not excluded

<TomB> Joakim: Some members want to see RDF implementation but still not excluded - would be like a slice

Joakim: Another common question: how does it relate to DC? At time DC was developed - there is no semantics of relationships to other attributes.
... Our attributes could be viewed as a vocabulary. We have attributes that are terms. Very easy to make ontology - the long-term purpose.
... We have semantic relationships in place. But group did not want to demand a technology in the implementation - would make it less useful.

<manu-work> http://connectedmediaexperience.org/technicaloverview.html

Manu: New standard being worked on that has strong overlap with Media Fragments - express music, television, movies, on Web in standard way - using RDF as base mechanism

Raphael: So this is basically a mapping and the working group is asking how to represent and whether it should be implemented somehow. Explicit RDF, SKOS, etc?

Joakim: Good summary.

<manu-work> Joakim, Raphael: my e-mail: msporny@digitalbazaar.com - follow up with me after the telecon.

Scott: In meantime, re: participating in W3C, in Health Care - seen other groups being created. Question: Specific apps that will help improve uptake of your work? Barriers to uptake?

Joakim: Two scenarios: one implemented in browser, another as Web service (if you are a metadata aggregator)

Scott: Do you see need for provenance for metadata?

<manu-work> Provenance shouldn't be their problem, right?

<manu-work> it's in a different layer?

Joakim: not mentioned in spec, but should be covered by different layer.

<manu-work> yes, that's right Tom.

Media Fragments WG Update

<raphael> My Slides: http://www.w3.org/2008/WebVideo/Fragments/talks/2010-04-07/

Raphael: Spoke to CG one year ago. 15 active participants.
... Goal: URI-based mechanism for identifying media fragments on Web.
... We spent several months on use cases.
... Slides 4 and 5 are user stories.
... People who want to share delimited sequences of videos and get URI to share via microblogging
... Lena wants to browse audio tracks of videos - not send all tracks over wire, but select tracks for display
... refer by name to particular chapters ...
... Slide 7: Media Fragment URI Syn tax for Time, Space, Track, and Name
... API in Javascript - browser can find out what tracks are available
... Track and Name are trickier, especially Int'ln problems
... Slide 8: Normally, hash is stripped out before sending to sever.
... will slightly change...
... we foresee smart agents that do not throw out hash, but encode into HTTP headers
... in order to serve fragments requested
... current browsers will just ignore - not break.
... Slide 9: http request, smart user agent can make mapping
... can expose response
... Slide 10: what is sent back is just the bytes corresponding to the sequence
... with HTML5 browser
... Slide 11: user needs server help. Custom units can be used in range requests.
... will extract bytes and send back with 206 partial content
... mapping between time and bytes, packages sent back to users.
... Slide 13: Optimization of Recipe 2. Question in seconds, answer in bytes.
... win cachability with this recipe
... Slide 15: optimization, different - request expressed in seconds, but serves multi-part response - not just bytes corresponding to sequence, but playable response

<ivan> worked for me, highlighting raphael:-)

Raphael: Slide 16: Demo. Click on Media Fragments player. Click on GO. Can just play sequence that is highlighted.
... click on Silvia's experiment with HTML5 and Javascript.

<ivan> that one does not work in safari:-(

Raphael: implementations in pipeline for HTML5
... normally should work with Safari, Opera, Chrome, Mozilla

Manu: Is this something we are expecting Apache and IIS teams to pick up and implement, or expecting modules that Apache uses? How do we expect this to be used in field?

Raphael: Currently talking with Squid (proxies) - hope to have Apache modules. Would be codec-dependent.
... would aim at good coverage of major formats.

<manu-work> This would be very useful for <video> and <audio> tags in HTML5.

Ivan: On client side, talking with browser vendors? Opera in WG, impelementing. Longer-term, will be native. Mozilla patch. Apple is in WG.
... Imagine it will be supported.

Raphael: We are in close touch with HTML5 working group.

Raphael: "Yes, will work with fragments".

Manu: Video and audio tags is primary element targeted?

Raphael: Targeting both, but any URI you put in browser bar. Flash clients that can also handle. Not just HTML5. But HTML5 really pushing.
... HTML5 covering understanding hash fragments.

Scott: QuickTime versus Flash? Do you see a problem?

Raphael: not really. Main problem with both: plug-in within browser. So if looking in browser environment, need communication with plug-in - security issues.
... Native audio and video HTML browsers handier to implement.

Ivan: URI structure will give additional structure for talking about fragments of media.

Meeting adjourned - thank you Raphael and Joakim.

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2010/04/07 14:18:35 $