PBS transcribes text to be search engine-friendly

PBS today launched a new feature, which allows users to search transcriptions, text and data from 7,000 free videos posted on PBS.org, the U.S. public television network announced.

The feature, powered by RAMP’s MediaCloud technology, will help users find, sort and share results, and filter by programme title, producer, local PBS station, airdate and content format. More than 2,000 hours of video from programmes are available, including top PBS titles like Frontline, Nova, American Experience and PBS Newshour.

Nieman Journalism Lab pointed out that online and in the search world, print publications have a huge advantage over video and audio – “they deal in words.” Video and audio, however, are practically invisible to search engines, which need text to crawl.

The automatic tagging’s accuracy rate is about 80 to 90 percent, which is very high in terms of automatic tagging, said Jon Brendsel, PBS’s vice president of product development, according to Nieman. RAMP will employ humans to review and refine auto-generated transcripts for a “hefty fee.”

Image: Search results for "dinosaurs" on PBS.org