Skip to main content

Finding themes in big data is still a matter of interpretation, says researcher

The 32nd annual School of Communication Van Zelst Lecture in Communication featured Paul DiMaggio, the A. Barton Hepburn Professor of Sociology and Public Affairs at Princeton University, whose lecture “Disciplining Interpretation: Machine-Learning Approaches to Studying Cultural Change” brought in a full house of faculty and student researchers from across the Northwestern University campus.

DiMaggio’s lecture, as well as a research leave given to a professor within the School of Communication each year, is made possible by the generosity of Louann Van Zelst (C49, GC51) and the late Theodore Van Zelst (McC45, GMcC48). Mrs. Van Zelst was in attendance on Wednesday, along with her children Jean Bierner (GSESP89) and David Van Zelst.

LouAnn Van Zelst (front) with her son David Van Zelst, DiMaggio, and daughter Jean Bierner.

DiMaggio was introduced by Eszter Hargittai, the April McClain-Delaney and John Delaney Research Professor in Communication Studies, a former student of DiMaggio.

“Machine learning” is the use of computer algorithms to find patterns in large amounts of complex data.

“These methods provide a way to identify analytic objects—themes—in large bodies of texts and to track their evolution over time,” DiMaggio said. “I’ve developed a new set of tools that enable us to take the output from such models and use them to identify the emergence of new frames or meaning structures, the competition or mutual support among rhetorical framings, and differences among groups in their interpretation of particular situations.”

DiMaggio developed the tools he presented to the group as part of a larger study of cultural contention in the U.S. during the 1980s and 1990s—the period of the so-called “culture wars”’—to complement work he had already done on the debate over federal grants to artists and arts organizations, including research on public opinion and studies of actual controversies over the arts and media. He and his colleagues analyzed newspaper coverage on government support of the arts, allowing them to spot trends over time in how the issue was framed and presented to the public, and how editorial policies, for instance, affected press coverage.

High-speed computing has given two gifts to social scientists who want to use rigorous methods to study cultural change, DiMaggio said.

“First, it has enabled digitization and rapid transmission of massive amounts of text; and, second, it has enabled the development of algorithms to analyze such texts in minutes or hours whereas, in the past, such analyses might have taken weeks or even months,” he said. “The result is that we can take ideas that, in the past, existed largely in the realm of theory and render them measurable and visible in ways not possible until recently. This capacity is creating a golden age of text analysis and revolutionary analyses in fields that study cultural change, from communications to cultural sociology to the digital humanities. At the same time, it is critical to recognize that the study of culture remains intrinsically and unavoidably interpretive.”

To use any of these methods well, the research team must include someone with deep expertise in the subject area, DiMaggio warned.

“For the subject-area expert, topic modeling and other computer-assisted forms of textual analysis can be very helpful in guiding and disciplining human interpretation, but can never be substituted for it,” he said.

DiMaggio has taught at Princeton since 1992. He serves as Director of Graduate Studies of the Department of Sociology and Director of the Center for the Study of Social Organization. Before coming to Princeton he taught for twelve years at Yale. DiMaggio’s published research on culture and the arts, economic sociology, organizations, inequality, political attitudes, and social networks has appeared in the American Journal of Sociology, the American Sociological Review, Media, Culture and Society, Poetics, and many other books and journals and has been translated into nine languages. A former fellow of the Russell Sage Foundation and of the Center for Advanced Study in the Behavioral Sciences, and a former Visiting Professor at the Annenberg School for Communication, DiMaggio is David Reisman Fellow of the American Academy of Political and Social Science and a member of the American Academy of Arts and Sciences.