Published September 20, 2012
From an ordinary computer, a Raytheon system can monitor television broadcasts from all over the world, 24 hours a day, 7 days a week -- and instantly translate and analyze them.
It’s amazing technology, stuff deployed operationally since 2004 by the Department of Defense for a variety of purposes, including open-source intelligence, information operations, public affairs and PsyOps.
Available through the General Services Administration, it does not require any hardware setup, software installation, or onsite administration or maintenance.
Could it have anticipated the escalating negative sentiment towards Americans linked to the foreign media coverage of the anti-Islam film “Innocence of Muslims” -- and perhaps helped protect U.S. embassies and interests abroad?
Raytheon’s BBN Broadcast Monitoring System automatically captures foreign media and deploys state of the art technology to translate, transcribe and analyze those video and audio streams in real time. It can spare a human some of the tedious and time consuming work by automatically sifting through the vast volume of foreign language news.
“Machines are good at looking for things and humans are good at analyzing … [it’s the] perfect bionic combination,” explained program director Prem Natarajan, head of speech and language processing at Raytheon BBN Technologies. “The machine does what it’s good at and the human does what it’s good at.”
Rather than the old school approach, analysts can use the Raytheon system to search topics, receiving back a list in chronological order of worldwide opinion as seen through media coverage.
Taking the recent backlash against Americans linked to “Innocence of Muslims” -- An amateur film made in the U.S. that has outraged the Muslim world -- technology like BBN Broadcast Monitoring could have been programmed to hunt for references to the inflammatory video in foreign media.
Using the system’s watchlists and alerts, it would automatically identify broadcasts referring to the video and therefore help build a picture of trending coverage and hotspots where violence against Americans might erupt.
On July 1, the video was first posted online; Arab TV stations in a number of countries soon began covering it. The religious Egyptian TV channel al-Nas broadcast scenes on Sept. 8, and a clip dubbed into Arabic was then posted online. Hundreds of thousands of viewers watched it in mere days.
Ultimately, four Americans including Ambassador J Christopher Stevens, were killed in an attack on the U.S. consulate in Libya. Reports have linked the attack to reaction to the video, as well as terrorist groups such as Al Qaeda.
On Monday, the Lebanese Shia movement’s Hezbollah leader appeared in public for the first time since December 2011 to denounce the video. Last Friday, Sunni militants in Tripoli, clashed with police and set fire to American fast food outlets.
How Does It Work?
On the computer screen, the user's view of the BBN Broadcast Monitoring System shows a transcript in both the original language and English. The system automatically transcribes the real-time audio stream and translates it into English, even punctuating appropriately.
Both the transcript and translation are searchable and synchronized to the video. It also automatically recognizes entities such as people and places, further enhancing accuracy and enabling advanced analysis.
The interface allows users to quickly search for specific spoken content in the video archive up to one year of broadcasts through keyword queries in English or the source language.
Search terms are highlighted in the results with an image thumbnail, the original language transcript and an English translation. Clicking on a search result jumps the user directly to the streaming video where the search term was mentioned.
Users can save searches in a “watchlist” that continuously monitors incoming video for matches to search criteria. The system automatically adds results to the watchlist as they are broadcast and alerts the user as new matches occur.
Watchlists can be filtered by date, user and topic, so users view only items of interest to them.
In the case of the anti-Islam video, the watchlist could hunt for references to “Innocence of Muslims” and identify where they appear.
The system is also very good at identifying relationships between individuals and building biographic or entity profiles.
Currently, ten languages are available. In addition to Spanish and French, these include some of the most difficult to master such as Modern Standard Arabic, Mandarin Chinese, Farsi, Bahasa Indonesia, Hindi and Urdu.
With this technology, users with no foreign language skills can get the gist of a broadcast and triage enormous volumes of media, freeing up skilled linguists to focus on other critical tasks.
Ballet dancer turned defense specialist Allison Barrie has traveled around the world covering the military, terrorism, weapons advancements and life on the front line. You can reach her at firstname.lastname@example.org or follow her on Twitter @Allison_Barrie