Mar 8, 2018

The Economist Explores David Bamman’s Algorithmic Literary Analysis

From The Economist

O fractions, my fractions! Machines are getting better at literary analysis

by J.T.

Algorithms that identify the voices of authors and characters should be celebrated, not scorned.

In “Dead Poets Society” (1989), John Keating, a teacher at a 1950s American boarding school, played by Robin Williams, draws a chart, its shape dictated by a fictional essay called “Understanding Poetry”....

Doubtless Mr Keating would have been dismayed to read “The Transformation of Gender in English-Language Fiction”, a paper published last month in the Journal of Cultural Analytics. The authors—Ted Underwood and Sabrina Lee of the University of Illinois, and David Bamman of the University of California, Berkeley—have trained a series of machine-learning models on a broad corpus of 104,000 works of fiction written between 1700 and 2010.... The algorithms they have trained on the data have allowed them to explore a range of gendered issues....

Across a wider sample, however, they can be deployed more confidently. A paper published by Mr Bamman in 2013 was able to identify character stereotypes from 42,000 Wikipedia film summaries, which clustered Batman with Jason Bourne and the Joker with Dracula. A follow-up in 2014 confirmed various literary theories about the similarities between characters in the novels of Charles Dickens and Jane Austen, among other writers....

Read more...

Last updated:

March 8, 2018