Good morning. Today I'll be telling you about "data mining in networks" and, more specifically, about how data mining techniques can be applied to problems in counter-terrorism. This presentation is based on work with my students and colleagues in the Knowledge Discovery Laboratory at the University of Massachusetts Amherst, and also on work I did when serving as an analyst at the Office of Technology Assessment, an analytical support agency of the United States Congress. During my last two years at OTA, from 1994 to 1995, I coauthored a study on using data mining techniques to find evidence of money laundering in large databases of financial transactions. The findings of that study led to my current research interests, and it also led me to create my research group at UMass.

There are two things I should make clear at the outset. First, I am not an expert in counter-terrorism. There are many remarkably smart folks both inside and outside government who know a great deal about this topic. I am not one of them. What I do know about is data mining, and how current data mining techniques do (and do not) apply to the types of data I have been told are central to counter-terrorism analysis. I have had opportunities to talk with analysts in the intelligence and law enforcement communities, both when I worked at OTA, and now that I am involved in several research programs for creating data analysis tools. The needs of these analysts are unusual and challenging, and that's what I'll tell you about today.

Second, my research is supported by the Defense Advanced Research Projects Agency and the National Science Foundation, but I do not speak for them. My views on these topics are my own.