Features

GPS, Q7: What new ideas you are working on and excited about? Do you see a start-up in your future?

Kleinberg: I'm continuing to collaborate with people in the social sciences, particularly economics and sociology, and I think the opportunities here continue to expand rapidly. There is a lot we can learn from these areas about building models of human populations -- particularly, the decisions people make and the population-level feedback effects that arise from these decisions. At the same time, there are deep questions about what form such models should take, and how the data should inform the development of models, and these are kinds of questions where the techniques we specialize in can be very useful.

It is also very interesting to build stronger connections between the types of network studies going on in data mining on the one hand and sociology on the other. The bulk of empirical network research in sociology has focused on relatively small systems (up to a few hundred individuals) for which researchers had a good understanding of the people being studied and the meaning of the connections between them. Now, when you go and collect on-line data -- say, from a social networking site -- so as to study a network with a few million people, you end up a very different situation. You know much less about who the people are and what the links mean; but at the same time, you can potentially identify patterns that, while genuine, are too subtle to be visible at smaller scales. Thanks to increasing communication between the disciplines, these approaches are moving closer together; and the ultimate goal, of course, is to find the point at which the two lines of research converge -- giving us the ability to ask complex, nuanced questions from the social sciences on social networks of enormous size.

In a slightly different direction, there are many interesting questions in the development of tools to mine one's own personal data. As we accumulate digital archives capturing the history of our personal reading, writing, listening, and communication habits, we're going to be increasingly interacting with tools that mine this in the course of shaping our on-line experience. And of course, one can't go very far down this line of thinking without running into fundamental questions about the privacy of this kind of data, a problem I've been thinking about recently with Lars Backstrom and Cynthia Dwork in the context of social networks.

(GPS: see L. Backstrom, C. Dwork, J. Kleinberg. Wherefore Art Thou R3579X? Anonymized Social Networks, Hidden Patterns, and Structural Steganography. Proc. 16th Intl. World Wide Web Conference, 2007.)

As for your question about start-ups -- that's of course a recurring issue for anyone working in this area. Personally, I like being in academia -- both the opportunity to interact and collaborate with people in very different disciplines, and the impact that comes from dealing with students at different levels. But part of my view is based on the fact that we're talking about an area in which the movement of ideas between academia and industry has been particularly fluid; people at both large and small companies are aware of what's going on in academic research, we all attend the same conferences, and there's a sense that you can do things that will really get used.

What's this?