Logo: University of Southern California

Viterbi Effort for CREATE Attempts to Bring Order to "What If"

"Risk Analysis Workbench" planned for use by all Homeland Security Centers

June 01, 2007 —

RAW Talent: Tatiana  Kichkalo, left, with project  leader Michael Orosz
A Viterbi School team working under the auspices of the Department of Homeland Security-funded Center for Risk and Economic Analysis of Terrorism Events (CREATE) is creating a system to help policymakers with a faster more consistent way of quantifying risk estimates.
Information Sciences Institute project leader Michael Orosz is leading the team working on the tool, called Risk Analysis Workbench, or RAW, which will ultimately be used at all eight DHS research centers, not just CREATE, which is located at the University of Southern California.

ISI is part of the USC Viterbi School of Engineering, which along with the USC School of Policy, Planning and Development are co-hosts of CREATE.

The project brings longstanding ISI strengths in web resources, information sharing and artificial intelligence to bear on the task of gathering and distributing the specific data necessary to perform the tree analyses and other problems of "what-if" risk analysis in a speedy and, above all,  a uniform way.

ISI insights and inventions around the "semantic web," -- systematic structuring of database information to make it more accessible over the Internet -- are a major part of the system, which is also incorporating efforts by ISI’s Robert Neches and Tatiana Kichkaylo.

RAW grows directly out of work by CREATE director Detlof von Winterfeldt, a professor in the Viterbi School department of industrial and systems engineering with a distinguished career stretching back decades in the field of risk assessment.

In an interview, Von Winterfeldt illustrated the way that the analysis proceeds by pointing to a specific CREATE study he himself completed, about the dangers posed by a small, man-portable anti- aircraft rockets being used to bring down a passenger airline, tracking consequences out a "decision trees" of alternate possibilities to evaluate costs and benefits.

The estimate for the cost of installing countermeasures, Winterfeldt says, is about $30 billion for the air fleet, which at first sight seemed completely out of line with the damage that an attack would cause. But tracking out the decision trees, more and more costs appeared growing out of a successful attack, widening impacts.

RAW data: Schematic diagram shows overall design of system, now in version 2.0.

The bottom line of the study was that the installation of the defenses would be cost effective if three conditions prevail: 1) the probability of a MANPADS attack is greater than 40 percent over 10 years; 2) the economic losses are very large (greater than $75 billion); and 3) the countermeasures are relatively inexpensive (less than $15 billion) to implement.

This analysis by Von Winterfeldt was a widely admired achievement of CREATE’s first years — but the intensity of effort required to reach the conclusions raised an obvious question. Risk assessments have historically been what the British call “one-off” cases – isolated single efforts attacking a single problem.

But the wide needs of Homeland Security in many areas for parallel products has led to the RAW effort, an integrated, generalized toolbox designed to be used by risk analysts all over the nation, at all the centers, in a standardized but flexible format.

RAW is still in development. The current version is RAW 2.0, which according to Orosz will be distributed in July an ‘alpha’ or research version, rather than a “beta” - ready for trial use in the field. As an example problem, the ISI team is now working to integrate into RAW risk analysis resources used in foot and mouth disease research being undertaken at one of the other centers, the Center for Foreign Animal and Zoonotic Disease Defense, (FAZD), located in Texas, and will be reaching out to the other six centers for subject matter.

The system, says Orosz, "is designed so that you don’t have to be a computer scientist to use it."

Orosz hopes to distribute a beta version of RAW later this year.