• Intro
• News
• Capabilities
• Publications
• Community
• Contact

Today’s asymmetric threat takes advantage of all possible sources to utilize, manipulate, and broadcast its message.  Terrorists recognize the influence the media has, so they frequently use their own media committees to determine how best to manipulate information. From an intelligence perspective, discovering this message is difficult given the overwhelming backdrop of unrelated stories and noise.  Complicating this situation are the different types of news media to be considered including text, audio, and video.

To address this issue we must develop technologies that enable military and intelligence analysts to quickly identify and discover events in the news media to support the overall analytical process.  These capabilities must provide the ability to:

• Discover new activities in the news that may otherwise go unnoticed.
• Provide insight into what the enemy is thinking.
• Extract and correlate seemingly unrelated news media.
• Work across all modes of media including text, audio, and video.
• Support many languages including: English, Arabic, Russian, Spanish, French, German, Greek, Norwegian, and Polish.

There are three problems in the current “search” approach to monitoring enemy activities through the news.  First, the burden of correlating all of this data is put squarely on the shoulders of the analyst.  It is his job to sift through the different modes and formats of data to find related bits of information and put them into a form that can be analyzed.  With the volume of data that is available and constantly being published on the Web, this job is impossible.  Second, because results are limited to hits based on keywords or phrases, news items that are related to a specific query may never be discovered.  Third, new activities may emerge in the news that the intelligence analyst is interested in but has no way to gain that global awareness.

Here at DAC we have been scratching our heads regarding these problems for sometime now and I thought I would share some of the advancements we have recently made. Most of you are aware of the capabilities developed with the BOBCAT product suite, but I wanted to bring you up to speed on our MainShip product for monitoring news broadcasts. Also, I thought you might be interested in how we are bringing these two products together.

Mainship
The Mainship approach provides a powerful approach to monitoring news broadcasts in any language as is illustrated in the below picture.  In this example of the Mainship output, we show the source video broadcast (top center), the speech to screen conversion in its native language (Arabic, left), the translated text (right), and keywords of interest from the broadcast (bottom center).

Mainship is a hardware- and software-based media monitoring solution with the ability to tag and archive several channels of video feeds and to rapidly provide a sequence of video clips based on user queries.  Our Speech-to-Screen (STS) component combines a multi-language speech recognition capability with translation tools to convert foreign language audio broadcasts to English text.  This application is currently being used by the CIA, the FBI, DIA, and ONI.

BOBCAT
As most of you are familiar with BOBCAT I won't go into all of the specific details. The BOBCAT approach provides a suite of capabilities for discovering themes and exploring relationships found in unstructured data.  BOBCAT goes beyond simple trend analysis by turning data into actionable intelligence that can be used to predict future events.  BOBCAT accomplishes this by identifying Themes and Networks, Predicting events, and tracking them over time.

 In this figure we can see the themes that have emerged over time as of the evening of August 6, 2008.  On August 4 we see discussions of terrorist activities in Iraq and India, a peak about a terror attack in China, followed by Olympic security concerns in Beijing.  This illustrates the causality one can observe in trends using a tool such as BOBCAT.  The discussions about security concerns are no doubt in response to the terror attack.  A non causal example is observed in the Guantanamo Bay Terror trial example.  We can see in midday August 6 there was discussion in the news about both that and the Karadzic pending trial.  When a verdict was reached later that day in the terror trial, those news articles formed their own theme and spiked as news activity increased.

To use the “needle in the haystack” analogy, BOBCAT is able to identify the number of haystacks in a data set and give each one a meaningful label.  The analyst is able to scan the themes and quickly determine what is important and what is not, leading to more focused analysis. 

The BOBCAT/Mainship Integrated Solution

In the below figure we illustrate our integrated BOBCAT/Mainship solution for total information dominance of unstructured news media.  This approach supports news media in any format: text, audio, or video.  We work through this figure from right to left starting with step 1.  In step (1), unstructured text news sources are being indexed and processed by both BOBCAT and Mainship.  In step (2) BOBCAT and Mainship execute their powerful analytic capabilities including automatic activity and relationship  identification, 24x7 monitoring, and speech to screen audio conversion on the unstructured news sources.  In step (3) we illustrate how an analyst can still execute a specific query to retrieve a relevant subset of the text, video, and audio data as is shown in step (4).  A differentiator is step (5), where the analyst will discover new activities from BOBCAT that would be missed using search technologies.

Jim
jim.nolan@dac.us

Subscribe To This Newsletter

Email: