ToTS & TiP

Increasing the transparency of online algorithms We have recently entered the era of "big data", where the online activities of billions of people are now routinely collected and analyzed. This explosion of data has led to the development of numerous algorithms for tasks as diverse as online content recommendations, dynamic pricing of goods, and prediction of criminal activity. However, external observers---including researchers, lawmakers, and regulators---typically have only limited visibility into such systems, as both the algorithm itself and the input data are typically considered proprietary. As a result, the increasingly popularity of these systems has brought up significant concerns about their fairness, transparency, and potential discrimination. In this talk, I discuss my group's recent work that aims to increase the transparency of these systems via online algorithmic auditing. We have developed techniques that allow an external observer to determine properties of the algorithms, such as the extent to which outputs vary between users and the most important input features used to generate outputs. I describe our results from applying our techniques to three different real-world systems: content personalization in Google search, price discrimination in popular e-commerce retailers, and the surge pricing algorithm in Uber. Overall, our results offer a first step towards increasing the transparency of big data algorithms. Speaker: Alan Mislove is an Associate Professor at the College of Computer and Information Science at Northeastern University. He received his Ph.D. from Rice University in 2009. Prof. Mislove’s research concerns distributed systems and networks, with a focus on using social networks to enhance the security, privacy, and efficiency of newly emerging systems. He is a recipient of an NSF CAREER Award (2011), and his work has been covered by the Wall Street Journal, the New York Times, and the CBS Evening News.