6. Extracting Events from Text and Grad School Memories with Brendan O'Connor and Brandon Stewart

Our guests in this episode are Brendan O'Connor, Associate Professor of Computer Science at UMass Amherst, and Brandon Stewart, Assistant Professor of Sociology at Princeton University. We talk with them about their 2013 ACL paper (with co-author Noah Smith) “Learning to Extract International Relations from Political Context” which presents a probabilistic model for extracting events between countries and international organizations from news articles. Brendan and Brandon also discuss how their collaboration grew from "saying nice things" about each other's work to 30-page written research memos sent back and forth. We also discuss the "ballooning and focusing" scope of research, clunky computer labs in the early 2000s, challenges in incentive structures for interdisciplinary collaborations, and data replicability standards.

Om Podcasten

Large-scale data has become a major component of research about human behavior and society. But how are interdisciplinary collaborations that use large-scale social data formed and maintained? What obstacles are encountered on the journey from idea conception to publication? In this podcast, we investigate these questions by probing the “research diaries” of scholars in computational social science and adjacent fields. We unmask the research process with the hope of normalizing the challenges of and increasing accessibility in academia. Music: Jon Gillick.