Found this in my inbox sent by email from Dr Strangelove who has a blog called ‘Watching YouTube’ on which he has collated over 270 academic articles on the video-sharing website. Included in this bibliography is a list of articles on tagging and taxonomy subjects related to YouTube. From the bibliography list:
Crane, Riley, and Didier Sornette. ‘Viral, Quality, and Junk Videos on YouTube: Separating Content From Noise in an Information-Rich Environment.’ In Proceedings of AAAI symposium on Social Information Processing, 2008.
Ding, Ying, Ioan Tom, Sin-Jae Kang, Zhixiong Zhang, and Michael Fried. ‘Mediating and Analyzing Social Data.’ In Proceedings, Part II, On the Move to Meaningful Internet Systems, OTM 2008 Confederated International Conferences, Monterrey, Mexico, 9–14 November 2008.
Geisler, Gary, and Sam Burns. ‘Tagging Video: Conventions and Strategies of the YouTube Community.’ In Proceedings of the 7th ACM/IEEE-CS Joint Conference on Digital Libraries, Vancouver, BC, 18–23 June 2007.
Lin, Wei-Hao, and Alexander Hauptmann. ‘Identifying Ideological Perspectives of Web Videos Using Folksonomies.’ Association for the Advancement of Artificial Intelligence Fall Symposium Series Papers, Menlo Park, CA, 2008.
Paolillo, John C. ‘Structure and Network in the YouTube Core.’ In Proceedings of the 41st Annual Hawaii International Conference on System Sciences, 7 January 2008.
Ulges, Adrian, Christian Schulze, Daniel Keysers, and Thomas M. Breuel. ‘A System That Learns to Tag Videos by Watching YouTube.’ In Lecture Notes in Computer Science, 415–24. Berlin: Springer, 2008.
A while ago Jay sent across some information on HTML 5 which has functionality to include a video tag. – W3 HTML 5 video overview He talked about testing Firefox 3.1 which has video tagging and embedding built in – 3.1 is still in beta.
Using audio and video in Firefox supports ogg and wav for the moment. All looks early days still as many attributes do not seem to be supported yet. But, other tests seem to be appearing around the fringes to get this moving.
I am fishing for more information to bring this together with this post ending up being a bunch of links and me needing more information from people immersed in this stuff.
Before GL turned up in Melbourne awhile ago for the mc workshop there was some excitement from kein.org about firefox supporting OGG Theora (opens source video format) which hit the video vortex list. On the videoblogging list Michael Verdi’s test with OGG on Firefox, a video screencast even. Upcoming presentation by Chris Blizzard on behalf of Mozilla at the open video conference in NY June 19-20. and a welcome and informative why open video? post by Blizzard (Jan 26, 2009)
There’s one exception to this: video on the web. Although videos are available on the web via sites like youtube, they don’t share the same democratized characteristics that have made the web vibrant and distributed.
In this video screencast demo on OGG in firefox the HTML 5 canvas function is revealed…no flash or QT “the user can interact with the webpage while the video is being displayed” April 15, 2009. This is out there and potentially opens up what has been sort after for sometime in earlier hypervideo research – dynamic access to points within the video while it is playing.
The traditional functioning of audiovisual archives is being completely reshaped by today’s technological advancements. The expansion of fast broadband networks and the availability of software, hardware and recording equipment have broken down the barriers to the production and distribution of audiovisual content. Large quantities of multimedia materials are flowing on the Internet and into the archives every day, and all over the world ambitious projects are set up to digitalise heritage collections. Moreover, media start to look more collective and inclusive: the ubiquitous “Web 2.0″ discourse promises new levels of participatory culture in which all users are producers, sharing, appropriating and remixing content, overcoming the old regime of top-down broadcast media. Blogs, wikis, social networks and “user-generated-content” tools are presented as the new wave of voluntary alliances that users seek online. Even the traditional media are swept away into the hype: the BBC designated 2005 as the “Year of the Digital Citizen”, in 2006 Time magazine chose “You” as the as its esteemed Person of the Year.
What did I pick up from the XML Melbourne Lab feedback?
The VD system was described as a “taxonomy of display.” A “recombinant video player.” There was confusion “Is it a content engine or a tag engine? Could we provide clearer context.
Another called it “Anti-TV…the opposite of YouTube…not a lot in the house” and an example that contradicts all the noise on the Internet through its slowness and stripped back minimal design.
The ability for a number of people in different locations to see varying news perspectives at the same time. The multi-window composition creates the opportunity for multiple perspectives to be viewed at the same time. The viewer can make their own judgments on that news item by engaging across a number of perspectives concurrently.
Also, there is the potential to respond to the concept of multi-tasking. Why aren’t there more systems that allow users to view multiple clips at the same time as a way of searching and deciding what they want to watch? Makes sense speeds up consumption. I think people are ready for this type of viewing but we are still locked into the security of one window viewing due to established cinematic and TV paradigms.
In a multi-window format with data increased a combination of stills and video could be used where photojournalistic type images come to life for short periods with audio overlays. An example www.mediastorm.org.
Overall, the multi-window aspect is what made the VD system unique. In comparsion, tagging is something that is growing fast around online video content.
Of course sport also came up. The slow cricket match playing in the left hand screen while other sports stream through other windows. Activities like sports can easily be watched at the same time as viewers wait for highlights the goal to be scored etc. The Olympics to die for in this system. Delayed edited broadcast another option, along with multi-camera curation except you see all the cameras. This tied in with live VJ gigs and music concerts.
Granularity – Semantic Video
Following up the idea of fragmenting existing TV programs for web publication the www.abc.net.au/fourcorners TV documentary program provides excerpts with duration times for viewers to access independently. But there seems to be a far as I can tell, no extras like out takes, extended interviews and other background. Also, the material seems to rely on previous program viewing with little focus on taxonomy, classifying under themes and categories.
A viewing platform with thumbnail similarities www.piclens.com a type of fly through viewing image-videowall but the clips remain separate as discrete independent pieces of content.
In a fast moving environment where time is of essence one person argued that time should not be invested in classification – taxonomy of online video content. The approach should be UGC instead where users make their own folksonomy type choices. An example is the vmark system. Here the idea is to leave long duration recordings and let users break the material up into whatever fragments they choose, with the option to embed and share those portions with themselves and others. (I need to try it out to confirm this perception) A Korean example of vmark – http://zzim.kbs.co.kr/section/ . A key objective is to get return traffic back to the original source material using metadata.
But, the concept of the content producer avoiding having to classify video content manually misses the point in relation to VD. Because the idea is to construct specific relationships between text and moving-imagery as way to provide certain types of context for the viewer/user.
A UGC idea where individuals capture material around Australai on the premise of classification rather than editing. These are single shots (no edits) but there could be jump cuts in camera, which are categorised and tagged. An approach that ties in easily with amateur online production techniques to shoot and publish directly online. (via a computer or direct from mobile etc) i.e. qik Many amateur producers struggle with more advanced editing but are becoming familiar with tagging and folksonomy practices. There could be themes where the content that is uploaded is synidcated into one central VD system for display under specific categories (themes that have been worked out in advance).
Re-mix is another consideration, particulary across multiple windows. Not only are users open to remix there own version (a standard single-window video) but also users could remix across multiple windows. It becomes more like a DJ turntable combining audio and vision from multiple sources at the same time.
Added social networking functions
Netvibes lets individuals assemble their favorite widgets, websites, blogs, email accounts, social networks, search engines, instant messengers, photos, videos, podcasts, and everything else they enjoy on the web – all in one place.
Some people loved the simplicity of the player (as it contradicted the visual overload of most web pages and players). Others where dying to get stuff back in there where the video component is supported with components that develop and maintain community. I discussed this earlier when comparing the motives behind Showinabox and how we stripped most of the web 2.0 blog functionalities out. View2gether is an example of a “social viewing platform” and freebase.
Returning to single window output
There was a number of people who wanted to be able to take away a traditional single edited video clip from the system as an option. This got me thinking about the divide we have created between the VD system and standard viewing practices. LIke creative commons currently considers the established status quo of traditional copyright until things move toward a more open approach. Maybe there is something in providing an additional single-window option. But a part of me also says NO, make the leap.
other reference – limelight networks
This application Sphinx-4 came up in discussions about tagging and annotating video content on the fly.
Sphinx-4 is a state-of-the-art speech recognition system written entirely in the JavaTM programming language. It was created via a joint collaboration between the Sphinx group at Carnegie Mellon University, Sun Microsystems Laboratories, Mitsubishi Electric Research Labs (MERL), and Hewlett Packard (HP), with contributions from the University of California at Santa Cruz (UCSC) and the Massachusetts Institute of Technology (MIT).
In the XMedia lab Nick Di Martino from The American Fim Institute (AFI) was first to demonstrate his wares, which included presenting a number of projects that are about to be released. He was quick to point out that “online video is exploding”.
Research funds seemed to be directed at very elaborate complex online video players. These players are attempting to integrate as many aspects of social media as possible into the interface. Another key issues is attracting users/viewers by optimising searching. Martino like many others in the conference stated that it is not worth trying to compete with meta-platforms like YouTube. The idea they argue is to distribute your content as far and wide as possible and look for follow-up or return traffic, along with channel partnerships with bigger platforms.
Martino demonstrated beta versions of the following AFI projects:
An issue for the AFI is working out how to integrate User Generated Content (UGC) with high-end production content and overall it is about balancing “scale, usability, uniqueness, ubiquity…” In this player UCG content is clearly delineated from professional production material. Later in the conference others spoke about creating an even approach to both forms of content where both are integrated and given the same sort of respect and hierarchy.
This player had elaborate ‘tag feed’ drop downs and a ‘tag manager’ that enabled users to add and manage their own tags. There was also a video cloud (a video egg) that used image thumbnails in a type of tag cloud structure. A ‘molecule’ feature worked like a VJ type application where molecules are connected in a hypertext type structure. These are portable like a widget and can be transferred to mobile hardware. Down the side was a community feed which acted as a forum for users.
A note worth thinking about in relation to thinking about chapters and tagging is planning linear content in advance so that it can be fragmented and classified.
Nic mentioned a semantic web type approach: Freebase about:
Freebase, created by Metaweb Technologies, is an open database of the world’s information. It’s built by the community and for the community – free for anyone to query, contribute to, build applications on top of, or integrate into their websites.
Already, Freebase covers millions of topics in hundreds of categories. Drawing from large open data sets like Wikipedia, MusicBrainz, and the SEC archives, it contains structured information on many popular topics, including movies, music, people and locations – all reconciled and freely available via an open API.
All this got me thinking about a design where categories are tags becoming a simple tag cloud like in the archives section.
The tags on the left and right become clip titles instead loosing a level. Categories are wiped. Categories in this interface are like making a batch or set in Flickr. The user selects content instead using the tag cloud directly.
I have got most of the clips into the ‘Glasshouse Birdman’ prototype and are now starting to look at how the categories and tags work in terms of how I would like the user to engage with the themes that have emerged in the content. I realised that clips could be kept in separate categories by controlling cross-overs with tag names, but sometimes a clip has something to offer in other categories. For example, giving a clip a tag name that features strongly in another category brings all those clips from that category across with it into the original category that has been selected. Often from my perspective this makes the theme to random if this is not the desired effect. One way around this within this interface design is to post the same clip twice with the same title but in a different category and with a different tag name.
category = birdman; tag = animal lover; clip title = big brown snake
category = feeding; tag = aviary; clip title = big brown snake
This means clips can be repeated to appear elsewhere while still having some control over themes. In a hidden kind of way clips that seemed more important that others could be repeated to appear in a number of places. Repetition becomes a feature of the narrative structure.
Chris let me know recently about a online documentary that he made titled ‘Thumb Candy’ on SMS text culture in the Philipphines that he put together within a blog. He gives Videodefunct a plug on the More about the project page as being an influence on using tagging and a blog to classify the video content.