期刊:Digital Scholarship in the Humanities [Oxford University Press] 日期:2024-09-25卷期号:39 (4): 1112-1122
标识
DOI:10.1093/llc/fqae053
摘要
Abstract Power relationships express one party’s dominance, control, influence, and authority over the other. In this article, and using state-of-the-art AI tools, we show that power relationships can be automatically identified in textual data. Generating thousands of synthetic utterances expressing either dominance or compliance, we trained/ran three models that showed good classification performance. Moreover, using GPT-4, we present a novel method for presenting power asymmetry in conversations and visualizing the dynamics of power relationships over time. This methodology is presented and illustrated by analyzing a case study—The play Pygmalion by George Bernard Show.