The Potential and Pitfalls of Topic Modeling for Humanities Research

This talk will introduce the text-mining technique called topic modeling, briefly explaining what it is and how it’s done. It will then turn to more substantial questions: what does this technique offer humanities researchers and what are its methodological limitations and problems? Both the potential and the pitfalls of topic modeling will be illustrated through research that uses topic models of newspapers to explore Civil War nationalism.

Dr. Robert K. Nelson is the director of the Digital Scholarship Lab at the University of Richmond. His current research uses a text-mining technique called topic modeling to uncover themes and reveal historical patterns in massive amounts of text from the Civil War era.  He is currently completing two projects from this research.  One is a digital project that will publish and analyze multiple topic models of Civil War-era archives including the Richmond Daily Dispatch and the New York Times.  The other is an essay that analyzes these models to produce a comparative analysis of Union and Confederate nationalism and patriotism.

This event is co-sponsored by the Institute for Advanced Technology in the Humanities, the Data Science Institute’s Center for the Study of Data and Knowledge, and the Scholars’ Lab.

This talk was recorded in Alderman Library, Rm 421 on February 25, 2015.  Click below to stream the podcast, and follow along with Rob’s slides.  If you’d like to hear more from the Scholars’ Lab, subscribe to our podcast series on iTunesU.

[podloveaudio src=”http://a643.phobos.apple.com/us/r30/CobaltPublic3/v4/02/76/7d/02767d82-95d8-2f30-6639-b9acaaa3f16a/304-7485002225613836972-nelson.mp3”]