Fine-Grained Detection of Solidarity for Women and Migrants in 155 Years of German Parliamentary Debates
Abstract: In this study, we evaluate the potential of large language models (LLMs), particularly GPT-4, for detecting fine-grained and detailed notions of solidarity and anti-solidarity towards women and migrants in German parliamentary debates spanning from 1867 to 2022. We evaluate the capabilities of recent LLMs on their ability to detect and categorize nuanced expressions of solidarity and anti-solidarity using a fine-grained social solidarity framework and apply the best-performing models to conduct a longitudinal analysis, aiming to detect and interpret long-term trends in political discourse. Our findings reveal significant shifts in the representation of solidarity and anti-solidarity, corresponding with historical events and changing societal attitudes. However, challenges remain, particularly in the model's sensitivity to the subtleties of political rhetoric and the limitations posed by partial dataset annotation.
Paper Type: Long
Research Area: Computational Social Science and Cultural Analytics
Research Area Keywords: Computational Social Science and Cultural Analytics, Human-Centered NLP, Multilingualism and Cross-Lingual NLP
Contribution Types: Data resources
Languages Studied: German
Submission Number: 869
Loading