Beyond YouTube: New Ways to Find Video on the Web
Wall Street Journal | October 30, 2008
Type "home improvement video" into a
traditional search engine and you're likely to get clips of the TV show starring
Tim Allen, how-to segments on lawn sprinklers and video of groundbreaking
ceremonies of a new Lowe's in
Traditional search engines depend on
video publishers to add tags and keywords -- called metadata -- to the clips
before they are uploaded to the Web. But a lot of videos lack detailed metadata,
making it hard for search engines to automatically categorize the content. Worse
yet, some videos may be tagged incorrectly -- sometimes intentionally, a
practice known in the industry as "tag spam." A video tagged with "Olympics"
that is actually a clip of a political attack ad is an
example.
So some video-search sites are
moving beyond tags and keywords. For example, San Mateo, Calif.-based VideoSurf
Inc. is using technology that can search the actual content of the video. Inside
YouTube, Google Inc. is experimenting
with speech-recognition technology to identify the words and phrases spoken in
videos. Some sites, like CastTV, comb the sites of publishers' such as CBS.com
and ComedyCentral.com and compile an index of links for one-stop surfing. Other
search engines are supplementing traditional search technology by mining for
additional details about video clips, such as how they have been rated by
viewers and how many times they have been viewed.
Web users in the
By far, YouTube is the leading host
of Web videos, garnering five billion video views in July alone, according to
comScore. Here, media companies as well as amateur videographers upload clips to
the site and insert metadata so that videos are searchable by type, by
popularity and by user ratings. But on YouTube, users will find only video that
has been uploaded to that site; they will miss millions of other professional
and amateur videos elsewhere on the Internet.
Tony Velasco, a 35-year-old
small-business consultant from
'Hit or Miss'
Search
For example, finding clips of
" 'Saturday Night Live' is very hit or miss" on YouTube, he says, because
oftentimes the content is pulled in a matter of days. So he must look
elsewhere.
That's why sites like CastTV search
for content directly from publishers, going beyond just YouTube. Alex Vikati and
her husband, Edwin Ong, founded CastTV after a trip to
Elsewhere, VideoSurf Inc. is
analyzing the actual visual content of videos using technology known as
"computer vision algorithms," which produces more relevant search results, says
Lior Delgo, the company's chief executive. Computer vision is the science of
programming computers to process and analyze images and
video.
For example, VideoSurf's technology
can identify characters within search results. A search for the television show
"Lost" brings up results for the show and also a thumbnail photo for each
character. Clicking on the thumbnail of "Lost" actress Evangeline Lilly will
bring up clips from the TV show and also other clips of Ms. Lilly, like her
appearance on the "Late Show With David Letterman." The site has indexed 10
million videos from 50 different online video
sources.
Google, meanwhile, recently released
a search widget called Google Elections Video Search. Using speech recognition,
videos uploaded to YouTube's Politicians Channels are transcribed and indexed.
This lets users search for words that are spoken in a clip. Viewer can also
automatically jump to the exact point in the video when a phrase is
said.
Several sites like Burbank,
Calif.-based startup Mefeedia incorporate social aspects -- like which clips are
making the rounds on popular blogs and what site uploaded the video -- into
their search approach. As a result, clips that have been rated highly by viewers
and that are embedded on popular sites are ranked higher in Mefeedia's search
results. The site has indexed 15 million videos from 15,000 sources, according
to the site.
Viewers are fickle, says Frank
Sinton, Mefeedia's chief executive. "Only one percent of people go to page two"
of the search results when looking for video, he says, so it's important to have
the relevant results on that first page.
Blinkx PLC, one of the bigger
video-search sites, is another company that has worked to improve its
understanding of viral videos, says Suranga Chandratillake, the company's chief
executive. Because it's difficult to determine which video clips will become
Internet sensations, Blinkx says it developed technology that points to clues
that a video will take off -- like if it gets linked to Wikipedia articles or on
social-networking sites. According to its site, blinkx has 350 media
partnerships and indexed 26 million hours of online
video.
Searching for TV
Shows
Blinkx has also expanded its search
services. In August, the company launched a new service called "blinkx Remote"
to search specifically for TV shows. Blinkx Remote indexes full-length clips
that are legally available from sites across the Web from sites like Hulu,
CBS.com and Showtime's site at Sho.com. The company hopes users will use its
site instead of visiting each site individually to find
content.
Human
Intelligence
Some sites haven't given up on human
intelligence when searching for video. OvGuide.com, a Los Angeles-based start
up, says it uses people to identify only the best video. "We only search what we
think is a good site so we control the results; and we also control the ranking
editorially," says Dale Bock, OvGuide's president and
founder.
OvGuide's staff reviews every Web
site that it searches to make sure only high-quality videos are on the site. If
a site has too many non-working video links or page after page of video spam,
the site won't be included in the search index. This approach ensures that
viewers can find videos that are relevant to them, he says. OvGuide searches
about 2,650 different sites and has more than 300 million accessible
videos.
Thad Mcllroy, a 52-year-old
electronic publishing analyst from
Mr. Mcllroy, who now watches a few
videos online a day, says he is looking for videos online more now. Before, it
was hard to cut through the clutter. "You get far too much thrown at you and
waste far too much time," he says.
> Back
