Name: Ronald Breiger –University of Arizona
Title: Toward Reconciling Big Data with Astute Thinking: Semi-automated Text Analysis of Documents on International Relations
Date: October 22, 2018 –MONDAY
Place: CASE 127
Abstract: Recent computational advances (such as natural language processing and topic modeling) allow massively huge amounts of text to be processed and patterns discovered while analyzing documents on international relations. These spectacularly efficient and highly useful pattern-detection methods do not, however, capture the nuance that a close textual reading would uncover, and in this sense the mechanized procedures are distant from the text. As between distant and close readings, we are currently exploring the possibility for a “third way” for text analysis: a transparent, low-tech text-reduction scheme based on how people actually use language. This low-tech approach involves locating a small number of text segments (hence reducing complexity) that can be read carefully by a human (hence preserving nuance). This talk will illustrate several tools for text analysis that enhance the control of human readers in the analysis of international relations texts, while still preserving many advantages of automated text analysis. To illustrate the benefits we are seeking from our “middle way” to text analysis, we apply this approach to public National Security Strategy statements issued by US Presidents in recent decades.
Short bio: Ronald Breiger is a Regents’ Professor at the University of Arizona, USA, where he is Professor of Sociology, Professor (by courtesy) of Government and Public Policy, and affiliated with the interdisciplinary PhD program in Statistics. His interests include theory and methods in social network analysis, the sociology of culture, organizations, stratification, theory and methods.