Intelligence analysts in the past have suffered from 'operational thrashing,' or too much data. Big data tools can help discover patterns and trends. ()
Big data tools and analytics have become essential for sorting through the growing mountain of intelligence data being collected from a rapidly multiplying number of human and machine-based sources
Whittling raw mounds of big data into actionable insights that can be rapidly disseminated to key users across the intelligence community (IC) and the military is a challenge that lies at the heart of ensuring continuous and reliable situational awareness.
Discovering the useful data
Whenever or wherever an important situation or mission exists, big data analytics can be used to uncover valuable nuggets of useful intelligence hidden inside mountains of structured, unstructured, real-time and legacy data assets. The trick to finding these knowledge fragments is to use the analytical tools that can reveal the greatest amount of useful information in the least amount of time. Once the intelligence has been retrieved, the next challenge is delivering critical insights to the IC and military users who can put them to immediate use.
“Big data platforms are the solutions that allow an agency to ask the right questions of data to gain intelligence into the actions of people or machines of any kind, essentially serving as the new intelligence platform,” said Mark Seward, public-sector senior director for San Francisco-based Splunk, an analytics software provider.
CAPT Christopher Page, deputy director of Assured Command and Control in the Office of the Deputy Chief of Naval Operations for Information Dominance, said big data’s potential has only begun to be tapped. “At present, big data is viewed a promising capability having the strong potential to broaden and deepen our understanding of the most pressing intelligence issues,” he said.
“Big data analytics provide fast recognition of patterns, trends and anomalies,” observed David Drake, technical director of the Communications and Information Directorate at the National Air and Space Intelligence Center. “This [technology] is well-suited to augment or even relieve human analytic burden when the area of interest is established and requires continual monitoring.”
“Big data analytics allow analysts to look broadly in new areas, such as parts of Africa, which are emerging areas of interest, and also to look deeply within familiar areas to understand what is happening — or what happened in the recent past,” said Lisa Shaler-Clark, deputy director, Futures, for the Army Intelligence and Security Command (INSCOM).
“Historically, the intelligence community ... has suffered from what is termed as ‘operational thrashing,’ where analysts are overwhelmed with the sheer volume of unstructured data, aged intelligence and circular reporting, making real-time accurate decision support extremely difficult, resulting in intelligence analysis paralysis,” said Peter Tran, senior director of the worldwide advanced cyber defense practice at RSA, a security consulting firm. “Big data application has improved signal-to-noise ratio efficiency, context enrichment, and integration of open-domain and closed-domain intelligence,” he added.
“Big data analytics provide significant benefits in discovery — the identification of patterns, trends and anomalies —which otherwise might be hidden in the noise of everyday events,” said Jon “Doc” Kimminau, Air Force analysis mission technical adviser for the deputy chief of staff for ISR. “These patterns and trends can richly inform analysis of the current situation.”
Big data’s strength is not identification of the current situation, it is discovering and understanding past activities and events, Kimminau added. “To the extent that the understanding of the past — previously not recognized — informs appreciation of current events, it supports and assists situational awareness.”
Pick data sources carefully
The effective management of sources has long been one of the key factors contributing to intelligence success or failure. That fact isn’t likely to change as the IC and military move deeper into big data. “Although some suggest that we can simply fill the ‘data lake’ with any and all forms of structured and unstructured data and then trust the results, there is still a need to ensure that we aren’t ... filling the reservoir with polluted water,” Page said.
Page noted the Navy works diligently to decide which structured and unstructured real-time and historical data sources about foreign naval forces and other entities are suitable for introduction into its analysis workflows. “The good thing is that we’ll be able to spend more of our time on making those critical decisions and less of our time and money on reformatting and other archaic data integration processes,” Page said.
The first key to big data success in situational awareness, Kimminau said, is for all sources to be made available in depth — not merely current situational data, but also data extending as far back in time as possible. The second key, he explained, is to include the observations and user notes as part of the data set — a sort of separate, additional “source.” “Without these two factors, data analytics are not really present,” Kimminau said.
“Sources of data that are literal in nature are the most well-suited for big data analytics,” Drake said. Textual data, whether structured or unstructured, as well as metadata are directly actionable, he noted. “Pictures and videos interpreted in a literal fashion — text transcribed, directly measurable, directly observable, etc. — may also be well suited,” Drake said.
Big data situational awareness analytics tend to fall flat, however, whenever sources are few or perhaps even nonexistent. “Some areas where situational awareness is needed are simply sparse areas where not much information is known by people outside the area,” Shaler-Clark said. “Remote areas or closed regimes are examples of these sparse zones where other [non big data-related] approaches may be more useful.”
The core of big data analysis: humans
Big data analytics, particularly when used to achieve situational awareness of rapidly changing domains, remains an inexact science. “One big challenge is that the data itself can have specific context and/or semantic meaning in relation to other data that may not have been accounted for during the creation of the big data analytic algorithms,” Drake said. “This could lead to ambiguous, missed or wrong conclusions unless the human analysts fully understand the assumptions and limitations when interpreting the results.”
Poor quality source data can also lead to inaccurate analyses. “Analytic judgment of the human analysts, who must remain vigilant, is essential to ensure successful use of big data analytics,” Drake said.
“It is more important to establish strategies that abstract tools, methodology and data than it is to select specific tools,” Drake added. “The goal is that tools and methods should come and go as needs arise, but well-managed data remains actionable, reusable and available.”
Paradoxical as it may sound, too much data, even high-quality insights, can actually be a drawback. “There will always be the problem of information overload,” said Alfredo Quiros, president and CEO of Telum, an operational support, analytics and situational-awareness technology provider. “A rifle-platoon leader doesn’t need live streams of information running through a HUD or earpiece as they are assaulting the objective,” he said. “This is where the human factor will always be needed to support and help filter out what is critical for the end-user to know on an immediate basis.”
Kimminau believes human analysts are the core of big data analytics and absolutely essential for situational analysis work. “We are only now defining the skills and competencies which make the best big data analysts,” he said. “The recent history of commercial big data indicates that data scientists and creative thinkers are the most needed.”
Page observed a need to strike an effective balance between “trusting the machine too much” and maintaining a healthy degree of skepticism about any answer that’s generated by a computer algorithm. “Having well-trained, well-led analysts in the loop is the best way to maintain that balance,” he said.
Shaler-Clark noted human analysts bring judgment, insight and curiosity to their work. “Their knowledge and experience guide the exploration of data to meet specific military needs,” she said. “Since the world keeps changing and bringing new challenges, the experienced human analysts bring their dynamic flexibility to build shared understanding.”
Big data is a tool that can be a force multiplier in terms of efficiency, but human analysts remain a vital part of making sense of the output and deciding courses of action, Drake said. Analysts also play a vital role in the continued development and refinement of the analytic algorithms that lie at the heart of big data tools. “Algorithms are optimal for forecasting known patterns, while analysts are vital for considering whole new types of data, use cases, and contexts not considered in the construction of the algorithms themselves, which is especially important in a dynamic time-sensitive environment.”
While big data analysts are critical to situational-awareness quality, such experts are in high demand across virtually all government and business sectors, making it hard for the IC and military to find qualified candidates. “That’s why it is so vitally important for us to continue placing heavy emphasis on recruiting, employing and retaining the best of the best,” Page said.