Breaking Down Big Data and Human Behavior

Dashun Wang
Physics, Ph.D.
Advisor: Prof. Albert Laszlo Barabasi
First Job: Research Staff Member, IBM TJ Watson Research Center, White Plains, NY

“When I started graduate school, we were witnessing a fundamental shift in the area of social systems — availability of large-scale datasets,” says Wang. “Indeed, just about everything we do leaves digital traces constantly being recorded in some database. Our whereabouts are saved by mobile phone companies for billing and routing purposes, and also by various mobile apps we installed on our smart phones. Whenever we make a purchase, what we shop for and our taste is indexed by credit card providers and the vendors we shop with. These large-scale datasets, capturing — in unprecedented detail — human activities, are expected to fundamentally alter our understanding of human behavior.”

Wang says his research has three main focus areas: “The first one is about the interplay between human mobility patterns and social networks,” he says. “How do an individual’s whereabouts affect who he interacts with, and vice versa.

The goal is to develop a mathematical and analytical understanding of the spatiotemporal aspects of human behavior.” The second focus, Wang says, is to understand human behavior in the face of emergencies and other extreme events. He says much progress has been made studying human dynamics during regular activities, but there is an exceptional need to also study human activity in times of great duress.

James Bagraw

“Lastly, and also more recently, we have been wondering whether there are any reproducible patterns behind success and future impact and if can we predict them,” he says.

Wang says, in one published paper, he and his fellow researchers asked if it is possible to predict social contacts by looking at where people have been — their mobility patterns. “We find mobility patterns reveal just as much, if not more, than state-of-the-art metrics about your friendships,” he says. “Most importantly, we demonstrate that when combining mobility and social network information, we have shockingly high confidence in predicting your next social contact, which also raises new privacy implications.”

In the case of emergencies, Wang says they’ve found that human behavior exhibits rather reproducible patterns. “We identified several patterns from three aspects: temporal (when do you call after an emergency and how long does it last), spatial (how far are you from the epicenter of the event), and social (whom do you contact during emergencies).”

Wang’s most recent research centers on understanding the laws that govern success. “We just finished a paper that quantifies ‘success’ in science — scientific papers,” he says. “The paper is currently in submission to Science*.”

Wang says he feels fortunate to have presented his, and his group’s, research in many international conferences and at some of the world’s leading institutes. “I found it a rather fascinating experience — to share with others my research and also to learn about others’ work,” he says.

Wang’s advisor, Prof. Barabási, says he’s been very impressed with Wang’s graduate work. “It is hard to be interdisciplinary, without compromising depth and quality,” Barabási says. “Dashun, as a student, has managed to find that middle ground. His research is farreaching, impacting many disciplines — from network science to computer and social sciences — without giving up the rigor of the physics perspective.”

*UPDATE: Wang’s paper was published in Science
D. Wang, C. Song, A.-L. Barabási
Quantifying Long-Term Scientific Impact
Science 342, 127-131 (2013). [ PDF ] [ Supplementary Materials 1 ] [ Fig(s). 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 ]
[ News & Views 1 , 2 , 3 , 4 , 5 ]

This entry was posted in Uncategorized. Bookmark the permalink. Both comments and trackbacks are currently closed.