Abstract: The vast quantities of language data that capture diverse aspects of human behavior offer tremendous opportunities for machine learning and language technologies to gain insights about society. In this talk, I show how natural language processing techniques can be expanded from understanding the meanings of words and sentences to inferring the underlying social structures and processes they reflect. In so doing, we can identify and intervene in crucial societal issues. I apply these techniques to computationally detect two ways in which the social context affects language use: 1) social relations affecting how individuals interact with one another, and 2) social constructs shaping how institutions interact with communities.
First, I show how to computationally detect various manifestations of power in workplace interactions between individuals — providing means to study the effects of incivility in those settings. Second, I show how to computationally investigate the ways race shapes the interactions between the police and the communities they serve — providing means for law enforcement to address and monitor racial disparities in policing. Finally, I talk about how these methods can be expanded to address broader issues such as incivility in online interactions. My research looks beyond words and phrases and introduces ways to extract the richer rhetorical and dialog aspects of interactions, and demonstrates the importance of such in-depth language processing for the computational social sciences.
Bio: Vinodkumar Prabhakaran is a postdoctoral fellow in the computer science department at Stanford University. His research falls in the interdisciplinary field of computational social sciences, with a focus on applying Natural Language Processing (NLP) for Social Good. He brings together NLP techniques, machine learning models, and social science methods to identify and address large scale societal issues such as racial bias and disparities, workplace incivility, and abusive behavior online. In his doctoral thesis, he studied how NLP techniques can help detect the underlying social power structures that guide social interactions. In his current research, he builds computational techniques to study racial disparities in various conversational aspects such as respect, politeness, and conversational structure within police interactions during traffic stops. His work has been published in computer science conferences such as ACL, NAACL, and EMNLP, as well as in high-impact multidisciplinary journals such as the Proceedings of the National Academy of Sciences (PNAS). His co-authored work on racial disparities in officer respect during traffic stops was awarded the prestigious PNAS Cozzarelli price for 2017.