Unveiling the Emotional Landscape of Ukrainian War Narratives using Large Language Models

Authors

  • Clara Maathuis Open University
  • Iddo Kerkhof Open University of the Netherlands

DOI:

https://doi.org/10.34190/icair.4.1.3060

Keywords:

Russia-Ukraine war, Generative AI, Large Language Models, LLMs

Abstract

Social media platforms serve as a dynamic environment where narratives, sentiments, and ideologies intermingle, catalysing the amplification and dissemination of ideas and experiences. As the ongoing endurance war in Ukraine represents a significant inflection point in the contemporary geopolitical landscape, revealing the urgency for comprehensive understanding and analysis of the implications and consequences of the activities therein. While traditional media outlets as we as research and practitioner efforts tackle various social, economic, and political dimensions of this conflict, efforts directed to capturing and reflecting on the nuanced thoughts and emotions of Ukrainian people are still incipient in relation to unconventional social media platforms like TikTok and Telegram. To address this gap, this research develops a novel LLMs (Large Language Models)-based modelling solution that captures, structures, and analyses the discourses characterizing the sentiments and emotions of Ukrainian Telegram users. This is done pursuing a Data Science methodological approach focusing on the insights collected in the first year of war, thus between February 24, 2022, up to February 23, 2023. By harnessing the analytical capabilities of LLMs, this research aims to bridge the gap between conventional understanding and the nuances of human sentiments, thereby advancing comprehension of the multifaceted dimensions of the ongoing war in Ukraine.

Downloads

Published

2024-12-04