Video Search: The Next Search Frontier

By March 26, 2007

Video has invaded the Net. Vbloggers, traditional broadcasters, independent directors and ordinary users are all contributing to the video glut. Video search engines are trying to rise to the challenge. A survey conducted in July-August 2006 for the Associated Press and AOL looked at how American online video viewers discovered videos. Only 37% used a search engine while 69% came across videos as they browsed the Web, 61% heard about videos from friends and 58% had a list of sites they visited regularly. Clearly, video search engines have their work cut out. Established Internet and search players, including AOL, Google and Yahoo, are of course offering their own solutions. But new actors are also emerging. Many of the video search engines share two features that have become commonplace in the Web 2.0 era: users can often upload their own creations and they can easily share videos they found with their friends.

The big three

Google Video. Not surprisingly, a search on Google Video brings in a lot of YouTube videos. On Google Video, you can watch videos for free. You can also buy or rent “premium content”. Navigation is intuitive, particularly to YouTube regulars.
AOL Video Search. Results are retrieved from traditional TV channels, MySpace, Yahoo or even YouTube (Google and AOL have been collaborating on video searching). AOL boosted its video search and video crawling capabilities by acquiring Truveo over a year ago. Purchasing TV episodes and other content is also an option here.
Yahoo Video. Nothing radically different on Yahoo Video which, like the first two, helps viewers share their findings with friends.

The contenders

Blinkx. This seems to be the engine to watch with persistent speculations that Google could be (or should be) interested in buying the specialized engine. Blinkx retrieves videos from a variety of sites. The list of results comes with a preview of each video. In its own words, San Francisco-based Blinkx is “fed by automatic spiders that crawl the web for audio video content and content partnerships with over 200 leading content and media companies” and it “uses visual analysis and speech recognition to better understand rich media content.”
Podzinger. A subsidiary of BBN Technologies, a speech recognition technology company, Podzinger uses that technology “to search words within both audio and video, not just the metadata, to classify content based on topic and usage.”
Dabble. “Dabble is about people describing, discovering and organizing video, wherever it’s found or hosted.” By creating playlists, users keep their video collections in one place and share them with others. The Berkeley-based search-cum-community site ranks videos on how often they show up in people’s channels or playlists as well as how much authority the channels or playlists have.
Also watch Pixsy which searches photos and videos; Quintura which mixes word-based queries with a cloud of tags to refine the search; which prides itself on letting users tag specific scenes within videos; Mamma, a master’s thesis project turned metasearch engine with an emphasis on video; TV Eyes whose searches focus on TV clips posted on the Web; CastTV, a yet-to-be-launched search engine which promises that its patent-pending technologies tackle the challenges of video search. Riya is concentrating on photo searching with the ambitious goal of learning to recognize faces once the user has identified them once for the engine.

Legal mentions © L’Atelier BNP Paribas