As rich media files, such as online videos, podcasts and widgets, become more common online, the next challenge is finding the best way to optimize them for search engines.
The first hurdle is to translate the content found in these files into text. The extracted keywords from those files play a crucial role in determining the relevancy of particular search query terms, said Deepali Tamhane, senior product manager of Yahoo Search, when reached by e-mail.
Speech recognition technology is one way to access the content of a given video or audio file quickly, and then translate it into text. However, speech recognition is not always an effective way to derive meaning and understanding from what is inside a video, said Alex Castro, CEO and co-founder of Pluggd, a video technology provider.
“We don’t think that a speech recognition approach by itself is going to be that effective,” he said.
For example, if someone is searching for “Tiger Woods” in a given video, he or she is not necessarily looking for every segment where the announcer mentions the golfer’s name, Castro said.
In order to help make video search more relevant for the users, Pluggd has developed semantic concept models that encapsulate meaning, Castro added. By using these models, statistical relationships can be established between words. Those statistical relationships can then be classified into categories like sports or finance or politics, he said.
As the ability to search within a video improves, users will be able to find relevant content faster and the amount of time they spend consuming that content will increase, Castro said.
“That ability translates into more potential revenue from advertising by the publishers,” he continued. “For advertisers, the benefit is being able to tightly [tie a] message to the content in a contextually relevant way.”
Once the video itself is searchable, then it’s merely a matter of time until video search across a given Web site and later across the entire Web can be optimized, Castro said. The first step, however, is cracking the code of search within a video, he said.
The explosion of rich media files on the Web has the power to revolutionize the Internet advertising world — as ads will no longer be textual links, but rather interactive videos and widgets that will add to users’ overall online experience, Tamhane speculated.
As rich media files continue to proliferate on the Web, Tamhane anticipates that click-through rates will no longer be as important to advertisers.
Instead, the focus will be on how much time users are spending on the site. But it should still be two or three more years before the technology is able to achieve this level of optimization for rich media files, such as video and audio content, Tamhane said.
“Delivering video content on the Web is still very new,” said Ketan Shah, VP of online marketing services for Spot Runner, an Internet-based ad agency. The necessary technology just isn’t there yet, he said. “Ithink it will take time.”
For now, Shah is advising his clients to optimize their Web pages where the video appears with keywords related to the content of that video.
