Indexing Web pages looks like child's play next to indexing video files -- the next big frontier for Internet search engines. But they're giving it a try.
Yahoo rolled out a video search service (video.search.yahoo.com) Wednesday that lets people look for videos online in the same way they hunt for Web pages -- by typing in a word or phrase and getting back a list of links to potentially relevant sites.
Another desktop search company, Blinkx, launched a more ambitious video-search engine on Thursday, dubbed BlinxTV (http://www.blinkx.tv ). It captures video and audio directly from Web sites, then uses special software to index these programs and make them searchable.
Rivals Google and Microsoft, meanwhile, also are working on multimedia search tools, while America Online already bought a video-search start-up called Singingfish.
Making video files searchable remains a largely unsolved challenge. "Video is hard to discover on the Internet," said Bradley Horowitz, director of multimedia search for Yahoo. "Web pages are self-describing. Video is opaque; it is a bucket of bits and doesn't tell you much about itself."
Yahoo's approach has mostly been to cooperate with video providers that can provide information about the movie trailers, music videos and other files they publish online. When it unveiled its video-search service, Yahoo also announced a new standard for describing media files, Media RSS, that it hopes others will adopt. This basically lets Web sites attach captions and credits to video and audio files that can be read and indexed by Web-search software.
Horowitz didn't say how many video files Yahoo's new service has indexed, but more than 5 million video files turned up in a "wild card" test search. The video search engine can locate files created in many formats, including Microsoft's Windows Media, Apple Computer's QuickTime, and RealNetworks' RealMedia.
Yahoo is not yet using software to index video files by automatically transcribing the words spoken in them, though it plans to do so at some point. Blinkx, however, is already taking that leap. This company, based in London and San Francisco, uses speech-recognition software to transcribe content from TV or video stored on the Internet, then uses that output to classify video that lacks some other description. The company said it has indexed tens of thousands of TV shows and other multimedia files hosted on the Web.
A new take on photo publishing
A photo-sharing Web service launched last week called HeyPix (www.heypix.com) blends Internet photo albums and Web logs.
Like several other picture-sharing sites, HeyPix (created by Arlington, Mass.-based WindUp Labs) attempts to add a social dimension to online photo albums. In addition to helping people display photos on their blogs by hosting them on HeyPix, the service offers a syndication feature that lets friends, family and even strangers subscribe to a user's public channel -- using any program that supports the popular RSS standard -- and be notified when the user publishes new pictures. Another sharing site launched this year, Flickr (www.flickr.com), also supports sharing over RSS, but most photo sites have yet to offer that feature.
The free version of HeyPix provides 50 megabytes of storage and accepts up to 50 photos. For $4.95 a month, those quotas get bumped up to 1 gigabyte and 1,000 photos; a $7.95 version offers 3 gigabytes of storage and up to 3,000 images.