Seminar: Predicting the Past; Using Machine Learning to Illuminate Three Centuries of Literary History - Ted Underwood (University of Illinois)
Predicting the Past; Using Machine Learning to Illuminate Three Centuries of Literary History
What can scholars do with large digital libraries? We're certainly comfortable searching and browsing them, and we're beginning to get used to the idea of mining patterns: we can visualise maps and networks and trends. On the other hand, interpreting large-scale patterns often remains a challenge. To address that problem, a number of literary scholars have begun to borrow methods of predictive modeling from machine learning. Instead of tracing a pattern and then speculating about what it means, these scholars start with a specific question they want to understand — for instance, how are men and women described differently in novels? Then they explore the question by testing models that make predictions about unlabeled examples. For instance, if we only know what a character does in a story, without names or pronouns attached, how easy is it to predict the character's grammatical gender? Since the past already happened, the point of making predictions about it is not really to be right. Instead we trace the transformation of cultural categories by observing how our models work, and where they go wrong. Professor Underwood describes how these methods have been used to illuminate the differentiation of genres, the emergence of distinctively literary diction, and the waxing and waning of gendered assumptions about character.
Contact firstname.lastname@example.org for more information about this event.