Speaker Duncan Watts, Fernando Diaz, Kate Crawford, David Rothschild, Justin Rao, Jennifer Chayes, Dan Huttenlocher, Kathy McKeown, danah boyd, and David Pennock
Affiliation Microsoft Research New York City, Microsoft Research New England and New York City, Cornell Tech, Columbia University
Host Jennifer Chayes
Date recorded 5 May 2014
In August 2013, Microsoft Research New York City moved into permanent lab space that was built-out specifically to meet the standards of the world-class research it needed to host. With strong ties to both the academic community and the technology industry, Microsoft Research is thrilled to open our doors in the heart of Silicon Alley, close to New York University and steps away from other Microsoft engineers in the Yammer and Skype divisions.
Duncan Watts - Computational Social Science: Exciting Progress and Future Directions
The past 15 years have witnessed a remarkable increase in both the scale and scope of social and behavioral data available to researchers. Over the same period, and driven by the same explosion in data, the study of social phenomena has increasingly become the province of computer scientists, physicists, and other “hard” scientists. Papers on social networks and related topics appear routinely in top science journals and computer science conferences; network science research centers and institutes are sprouting up at top universities; and funding agencies from DARPA to NSF have moved quickly to embrace what is being called “computational social science.” Against these exciting developments stands a stubborn fact: in spite of many thousands of published papers, there’s been surprisingly little progress on the “big” questions that motivated the field of computational social science—questions concerning systemic risk in financial systems, problem solving in complex organizations, and the dynamics of epidemics or social movements, among others. Of the many reasons for this state of affairs, I concentrate here on three. First, social science problems are almost always more difficult than they seem. Second, the data required to address many problems of interest to social scientists remain difficult to assemble. And third, thorough exploration of complex social problems often requires the complementary application of multiple research traditions—statistical modeling and simulation, social and economic theory, lab experiments, surveys, ethnographic fieldwork, historical or archival research, and practical experience—many of which will be unfamiliar to any one researcher. In addition to explaining the particulars of these challenges, I sketch out some ideas for addressing them.
Fernando Diaz and Kate Crawford - Social and Technical Challenges in Crisis Informatics
Over more than fifty years, information retrieval research has established a set of design principles which have been used to build information access tools for collections of legal documents, news archives, and even the Web. Crisis informatics refers to the study and development of information access tools for support during unexpected crisis events such as natural disasters and other human tragedies. These events often undermine many of the assumptions made in information retrieval research, resulting in system underperformance and catastrophic failure. We will present our approach to crisis informatics, which balances the insights from qualitative and ethnographic methodologies with engineering based on data-driven experimentation. By bringing together techniques from information retrieval and qualitative social science together, we can develop more robust and critical approaches that account for algorithmic challenges and the need for local knowledge and community engagement. This allows us to avoid biased results, signal gaps, and poor policy decisions. This talk will draw on collaborative work across MSR-NYC, MSRNE, MSR Israel and MSR Cambridge UK, including Javed Aslam, Matthew Ekstrand-Abueg, Megan Finn, Qi Guo, Virgil Pavlu, Soren Preibusch, Tetsuya Sakai, and Elad Yom-Tov.
David Rothschild and Justin Rao - Medium Data: Where Small Data Meets Big Data
People increasingly consume live events with second screens distracting them; in the near future this attention will be captured by the broadcaster. The broadcaster will display information, derived from new data analytics (econometrics and machine learning) on a mix of new and traditional data streams (fundamentals, social media, online, polling, and prediction games) with the goal of increasing interaction. The raw data from interactions is processed and fed back to the broadcast leading to more information and then more interactions. All of this happens in real time with increasingly sophisticated data infrastructure. While this is going to appear to the consumer as a “big data” revolution, in the near term, conventionally titled “big data” is going to be more of sideshow as the core revolution in low latency, quantifiable, personalized experiences is driven by “medium data”: faster infrastructure and more sophisticated analytics on traditional data sources.
Jennifer Chayes, Dan Huttenlocher, Kathleen McKeown, and Clay Shirky - Panel: Interaction Between the Academy and Silicon Alley in the Age of Data Science
Data Science, broadly defined, promises to advance many aspects of science and technology through personalization of everything from the way we interact with our phones to our medical treatments. In this panel, we discuss the promises and challenges of data science, as well as the responsibility of educating a new generation of data scientists.
©2014 Microsoft Corporation. All rights reserved.