Past Event

Amit Sharma – How Information Spreads in Social Networks: A Case Study on Prediction, Explanation and Intervention

April 5, 2018
2:00 PM - 3:00 PM
Event time is displayed in your time zone.
Computer Science Conference Room, Mudd Engineering Building, Room 453, New York

Speaker: Amit Sharma, Microsoft Research

Can an algorithm predict whether a tweet, photo or news piece will become popular? If it could, does it help us understand why some items become popular, while others do not? This simple question on information diffusion illustrates the tension between prediction and explanation in social systems: higher prediction accuracy may not necessarily yield good explanations, and convincing explanations often do not predict well. Using data from five different social networking platforms, Sharma will show examples of predictive models that achieve near-perfect accuracy, but tell us little about how content spreads. When we do try to design studies for explanation, common explanations fail to generalize. These results suggest that we are far from predicting or understanding what makes something popular besides simple temporal effects or “rich-get-richer” phenomena, and further that it may be impossible to do both. To resolve the dilemma between prediction and explanation, Sharma propose an approach based on causal inference that emphasizes designed data collection and continuous intervention. As an example, Sharma will describe an ongoing project on spreading mass awareness, Learn2Earn, that aims to understand the role of incentives in social sharing.

Free and open to the public.

Amit Sharma is a researcher at Microsoft Research India. His research focuses on understanding the underlying mechanisms that shape people’s activities as they interact with algorithmic systems, with an emphasis on the effect of recommendation systems and social influence. More generally, his work contributes to methods for causal inference from large-scale data, combining principles from Bayesian graphical models, data mining and machine learning. He completed his Ph.D. in computer science at Cornell University. He has received a Best Paper Honorable Mention Award at the 2016 ACM Conference on Computer Supported Cooperative Work and Social Computing (CSCW), the 2012 Yahoo! Key Scientific Challenges Award and the 2009 Honda Young Engineer and Scientist Award.

This event is sponsored by the Department of Computer Science at Columbia University.