Abstract: Today, there is more cultural data than ever before—more computationally tractable information related to literature, history, art, music, movies, video games, and more. This data has the potential to significantly advance our understanding of human culture, but not until we address significant technical, ethical, and disciplinary challenges—such as the methodological mismatch between humanistic and data-driven approaches or the frequent exploitation of user data across computing fields. In this talk, Dr. Walsh will share research that demonstrates how we can address some of these challenges by using computational methods and internet data to gain meaningful insights about cultural production while also respecting and protecting the amateur critics, artists, and everyday internet users who create most of this data. She will focus in particular on research that examines how users on Goodreads—the largest social network for readers on the internet and a subsidiary of Amazon—define, discuss, and debate “classic” literature. Analyzing more than 120,000 Goodreads reviews using close reading and natural language processing techniques like topic modeling, Walsh will show how thousands of Goodreads users collaboratively reimagine and reinscribe the “classics,” a category that is often thought to be defined by the opinions of cultural authority figures like professors and critics. She makes the case that computational methods and internet data, when combined thoughtfully, can enable researchers to capture the creative explosion of culture on the internet while also critiquing the most detrimental effects of data and computation on culture.

Bio: Dr. Melanie Walsh is an Assistant Teaching Professor in the Information School at the University of Washington. Previously, she was a Postdoctoral Associate in Information Science at Cornell University, where she worked with David Mimno. She received her PhD in English Literature from Washington University in St. Louis, where she specialized in computational approaches to text and social media data. Her research has been published or is forthcoming in venues including CSCW, the Journal of Cultural Analytics, Post45, Debates in the Digital Humanities 2023, and American Quarterly. She is co-PI of the NEH-funded “BERT For Humanists” project, which makes advances in artificial intelligence, like the large language models BERT and GPT-3, accessible to researchers in the humanities so they can better use, understand, and critique them. She is also co-editor of the Post45 Data Collective, a peer-reviewed, open-access repository for literary and cultural data after 1945.