The New Science of Visual Analytics
Over his 30 year career, Jim Thomas has specialized in the research and implementation of tools in scientific visualization, multimedia, and human computer interaction designed to increase the bandwidth of the connection between humans and the data they are trying to understand. Thomas has chaired graphics and visualization conferences for both ACM and IEEE, is the former editor-in-chief for IEEE Computer Graphics and Applications, and is on science and technology boards for universities, states, and industry. In short, Thomas has a passion for helping people understand what’s in their data.
But his most recent effort goes beyond creating new techniques within old disciplines — with his work in visual analytics, Thomas, Director of the National Visualization and Analytics Center (NVAC), has helped launched a new science that engages “…the human mind in a way that’s fundamentally different from the way we have done things,” as he puts it.
What is visual analytics? According to Thomas, “Visual analytics is the science of analytical reasoning facilitated by interactive visual interfaces.” It is traditional scientific visualization, taken to the next level. In “Challenges in Visual Data Analysis” (2006), Keim and his co-authors note that visual analytics combines visualization with human factors, information, geospatial, and scientific analytics. As pointed out in that paper, “Especially human factors (e.g., interaction, cognition, perception, collaboration, presentation, and dissemination) play a key role in the communication between human and computer, as well as in the decisionmaking process.”
An example? Consider the news stream at CNN.com and imagine that you are an analyst employed by Amalgamated Office Buildings to monitor what your competition, ACME Builders, is up to. You might be interested in news about big real estate deals, large purchases of steel, new metropolitan construction projects, and so on. But ACME is smart, and they know not to let news of their future projects into the press before they have the contract. There are clues, and any of these news stories individually might give a hint about what your competition is doing. The whole story, however, won’t be clear until you put all the pieces together.
So your job, as Amalgamated’s analyst, is to find the stories that reveal clues to ACME’s strategy in CNN’s massive news stream, and then put those clues together in a way that reveals what they are up to. Since you don’t know anything in advance, however, this is a lot like piecing together a puzzle where not only do you not know what the final picture looks like, you don’t even know where the pieces are.
Oh, and by the way, the news streams at NBC.com, ABCnews.com, and thousands of other news sites, blogs, podcasts, and videos on the Web have other clues that you’ll need to get an accurate picture. Good luck!
This kind of example is repeated countless times in border security, customs control, national security, disaster planning, crisis management, and so on. And this is the kind of problem that visual analytics in meant to solve. By using advanced computer-based analysis, graphics, and human-computer interaction techniques, visual analytics helps people sift through mountains of information to find individual pieces of a picture, and then helps them put that picture together.
One of the organizing forces for visual analytics activity in the United States and internationally is the National Visualization and Analytics Center, sponsored by the US Department of Homeland Security and headed up by Jim Thomas. The center, housed at Pacific Northwest National Laboratory and started in 2004, was created to develop and implement new techniques to help analysts, decision makers, and crisis responders manage the large volumes of unstructured information that can help maintain the security of the United States and its infrastructure in the face of potential threats and disasters.
The center has a far reach and broad scope, with both research and applied sides. The overall mission of the NVAC is described in its five primary charters: reduce the threat of terrorism; understand the risks to and vulnerabilities of our critical infrastructure; develop a visual communication infrastructure for response teams; create effective communications metaphors that can capture not only the conclusions of risk assessments, but also the evidence and chain of logic; and create an enduring talent base.
According to Thomas, about half of NVAC’s funding goes to basic research through five university-led research centers. Each center has a regional partner that helps to focus its work: Stanford, partnered with the regional FBI office; the University of North Carolina at Charlotte, partnered with the Bank of America; Purdue University, partnered with the Indiana University School of Medicine; Pennsylvania State University, partnered with the New York Port Authority; and the University of Washington, partnered with institutions around the Pacific rim. Each of these university research centers works on issues of specific interest to its partners; collectively, they advance the state of the art for the entire discipline.
The rest of NVAC’s efforts are directed to creating the techniques, and fielding the products, to help the nation’s infrastructure protection and response teams prevent, predict and manage crises. These products are where the rubber meets the road for what Thomas describes as “the new science of interaction,” with sophisticated reasoning and decision support tools wrapped in interfaces that are designed to be “walk up usable.”
Customers for NVAC’s products include the customs officials managing the stream of goods and people across our borders, intelligence analysts mining the rich but noisy streams of information on the Web, infrastructure protection teams, first responders managing a crisis as it unfolds, and many others. For these audiences, Thomas says it is critical to display the information in a context that is relevant to individual users. For example, professional analysts are accustomed to dealing with news flows represented as a visual landscape, with thematic areas shown as hills and mountains that grow and shrink with the volume of coverage in those areas. Power grid operators on the other hand, benefit from overlaying information about the grid on a geographical representation that puts the data in its physical location.
However it is not just the visual display but the interaction, or the “discourser of discovery,” as Thomas puts it, that enables discovery, understanding and confirmation.
Given all the attention on national security in NVAC’s mission, you might imagine that a good bit of what they do will remain trapped under a classified sticker for most of its useful life. Not so, says Thomas. There are about 200 projects in the US sponsored by NVAC, and publishing is encouraged in the open literature and in the six international conferences on visual analytics that happen every year. “Only about 5 percent of what we do is classified,” says Thomas.
When it comes to catalyzing growth in this new field, NVAC and its broad team of partners at PNNL, government agencies, academia, and industry are clearly accomplishing their goals. Thomas says that the field has grown from 40 researchers just a few years ago to over 1,000 researchers today, from all over the world. NVAC is also a “strong business area” for PNNL, bringing in roughly $25 million over four years. But these successes are only part of the picture. Through summer camps, research partnerships, and educational outreach, the center is looking out decades into the future to ensure that there is a talent base over the long term, helping to advance and refine this field. With no end in sight to the growth of data available in our society, these researchers are going to be sorely needed.