Is Google The Ultimate Context Miner, Refiner, and Producer?

August 08 2008 / by justinelee / In association with Future Blogger.net
Category: Government   Year: General   Rating: 4

This past June, Google-owned YouTube launched a new way to search for political videos on its YouChoose page:

Using speech recognition technology, the new function allows users to search for videos based on keywords that are spoken in the video. The resulting videos include yellow markers on the play bar to indicate where the keyword is uttered inviting the user to jump to that spot in the video. And if the user mouses over the highlighted area, a small overlay pops up with the phrase that includes the keyword, to provide some context.

Some critics have complained that providing only snippets of the entire transcript promotes people’s reliance on buzzwords to be informed and can result in information being taken out of context.

Currently, Google is testing the application’s success to determine whether or not it should introduce it to other parts of the site.

Google wasn’t the first to put this speech-to-text technology to use. Post 9/11, American defense contractors have been working on mining through online videos for possible terrorist threats.

Other video search engines such as Blinkx, Everyzing and Delve are also accessing the technology.

If users take a liking to the function, Google will very likely use it in their video advertising platform.

For example, instead of buying the current InVideo unit that includes an overlay advertisement that appears 15 seconds into a sponsored video, organizations can choose the exact location for the overlay advertisement to appear, based on keyword popularity.

Google must continue to improve its speech recognition technology in order to encompass video with speech that is not as straightforward. This may require some human computation.

But once the technology is perfected we should see a massive increase in the amount of valuable information at our fingertips.

Comment Thread (2 Responses)

  1. It’s just weird to think that speeches can be broken down into individual sentences, making the entire speech useless. The beauty of a good speech is the build up to a certain point, if you take away that build up then speeches might as well just be sentences read one after the other that have no connection to each other. You could say in one sentence something about health care, and the very next sentence something about abortion. Where’s the fun in that?

    Posted by: John Heylin   August 09, 2008
    Vote for this comment - Recommend

  2. I think this is good technology that will only enable people to find speeches they are looking for or about issues they care about—in the same way search engines have allowed us to search for websites with information we need. It doesn’t mean we will skim everything any more than we already skim written articles.

    I did just do a test of the technology on YouTube and it did well for single words, but not so well for phrases. There is still room for improvement.

    Posted by: Mielle Sullivan   August 11, 2008
    Vote for this comment - Recommend