For years now, millions of people share details about their real-world experiences on social media, including Twitter, LinkedIn, and Reddit. Using causal inference methods, we can data mine these social media streams to better understand the common and critical situations people are in, the actions they take, and their outcomes. Such inferences may be useful for many applications including decision support tools for individuals and analytics to support policy-makers and scientists. Over the last several years, we have performed several studies across a wide-variety of domains, including health, exercise, education and careers. In this talk, I describe some of the challenges and promises of extracting and interpreting potential causal relationships in the conversational space of social media, and the growing landscape of research in causal inference over social media datasets.
Emre Kiciman is a Principal Researcher at Microsoft Research AI. His current work focuses on causal analysis of large-scale social media timelines. Emre’s past research projects include social computing and search technologies, deployed in Bing; and foundational work applying machine learning to fault management in large-scale internet services, now an industry standard practice. He has over 50 publications, and 18 granted patents. Emre received his PhD in Computer Science from Stanford University, and BS in Electrical Engineering and Computer Science from UC Berkeley.