Can LLMs Extract Social Semantic Networks From Text?
Abstract:
Texts from books, news, and online media record social relationships and discourse of enormous interest for the social sciences. Advances in large language models have exciting promise to automatically infer social patterns from text, but questions around LLMs' robustness, transparency, and openness may challenge their validity as a social measurement tool. To explore some of these issues, this talk will describe a few recent and ongoing projects evaluating the abilities of prompted, low-supervision LLM methods to extract social semantic structures from text, including social networks in historical books, a media attention network on global events, and an argument conceptual network about vaccines.
Brendan O’Connor (http://brenocon.com) is an associate professor in the College of Information and Computer Sciences at the University of Massachusetts Amherst, who works in the intersection of computational social science and natural language processing – studying how social factors influence language technologies, and how to better understand social trends with automated text analysis. His work has been supported by the National Science Foundation, Google, Meta, and NSF CAREER. At UMass, he is an Associate Director of the Computational Social Science Institute. He completed his PhD at Carnegie Mellon University's Machine Learning Department, and BS/MS at Stanford in Symbolic Systems (cognitive science).
