This is a blog version of my magazine article for Inside Film, Issue #115. It’s a review and point of view on some of the latest developments in search and the importance of making rich content stand out online.

Searchable Video

Online video, whether it resides on Meta Cafe, Vimeo, Revver, MySpace, Current TV, or the goliath of YouTube, needs to be able to be discovered.

To put it into perspective, from a search query on YouTube in April, 83.4 million videos and 3.75 million user channels were returned. It’s estimated that last year, YouTube consumed as much bandwidth as the entire internet in 2000, and now around 13 hours of video are uploaded every minute.

Primarily our social connections help us to navigate and find content that we like. That could be via the friendly “check this out” emailed link, or a mixture of looking up favourites, or following and exploring related clips within the video channel itself.

But as the content grows exponentially in line with broadband uptake, lowering costs of equipment, knowledge of video editing and so on, a prized film could easily be lost in the noise of everyone else’s efforts. Despite the need to be illustrative in describing visual content – with informative titles, word tags and descriptions to help search engines – that very same search engine needs to come of age too.

In September, Google made headway in this area with the re-launch of Picasa, its online photo-sharing product – now with facial recognition technology thanks to its acquisition of Neven Vision, a company that specialises in matching facial detail with images on a centralised database. Picasa’s facial recognition technology works in much the same way. Users teach the database who’s who within a photograph and, over time, it will begin to suggest who is in an uploaded photo.

This isn’t moving image, but you can see how this technology could start to identify actors within a piece of footage, the scene they’re acting out and the film in which they’re starring. Apply this technology to voice-recognition software and you start to get the picture.

It’s early days and, even with photography, the technology works best when a person is facing the camera (it’ll have trouble identifying them if they’re not).

We still need to teach computers to understand faces and features, images and scenes. This is no small task – but can be made far simpler if it’s distributed among millions of web users.

The Amazon Mechanical Turk is one such service that divvies up human intelligence tasks among thousands of people. It creates a crowd-sourcing marketplace that enables computer programs to co-ordinate the use of humans to perform tasks that computers can’t – such as identifying emotions within photographs, a happy dog versus an angry one, or writing a description about an image. Users of the site can select from a number of different tasks and actually earn micro-payments from completing them.

A newly launched search tool named co5TARS is the latest in an emerging generation of content browsers that seek to address the increasing complexity of the web by visualising it. Unlike other solutions, co5TARS has been purpose-built to tell a particular story. Using the open data freely available from Freebase.com, it tells the stories behind the people in movies.

In a way co5TARS is a social network of film, depicting the roles and relationships of movie professionals as they change over time, offering unique views of this ever-changing network of actors, directors, writers, cinematographers, editors and producers, and using movie posters and iconography as a visual shorthand to navigate and explore.

Cooliris, formerly known as PicLens, is a web browser plugin that provides interactive full-screen slideshows of online visual content. The plugin, once installed, provides a rich ocular interface to explore content visually. Searching is done as you would with any content browser, by keywords, but the results are entirely visual and are displayed in a simulated three-dimensional interface for exploration.

It’s apparent that we’re ready for intelligent ways to find what we’re looking for. This evolution is being dubbed the semantic web – it’s the way in which information and services on the web are defined in a semi-automatic way, making it possible for the web to comprehend and satisfy the requests of people looking for content. The semantic web is the vision of Sir Tim Berners-Lee, credited for inventing the World Wide Web, as a universal medium for data, information and knowledge exchange. It’s also our want to have a smarter relationship with the web, so that when we search for Paris, the results are smart enough to know that probably we mean Paris France not Paris Hilton because it understands our previous behaviour as a traveller and not a celebrity gossip monger.

Users don’t care how the information sought is delivered. But for content creators it pays dividends to identify and be descriptive about the content being made and published online.