Automating Protest Event Analysis Using Digital Media in Contentious Politics addresses one of the most critical challenges in the field: the need for more efficient, reliable, and scalable data collection. Using Russia as a primary case study, author Bogdan Mamaev explores how Large Language Models (LLMs) and digital media are reshaping the methodology of protest event analysis (PEA).
Investigating the transformative impact of Generative AI on data access and efficiency, Mamaev argues that state-of-the-art proprietary and open models address the cost and resource constraints of traditional manual and semi-automated approaches to PEA, enabling the creation of high-quality datasets. By employing techniques such as zero-shot classification, Named Entity Recognition (NER), and semantic deduplication, researchers can extract rigorous quantitative and qualitative data from various sources, including news archives and social media platforms. Focusing on Russia, this work explores the complexities of building the Russian Contentious Events Dataset for News and Social Media (RCED-NSM) within an authoritarian regime characterised by heavy censorship. The book also examines the limitations and biases of algorithmic tools, testing their generalisability through comparative applications in China and the United Kingdom.
A pathbreaking contribution, this timely advancement in computational social science bridges interdisciplinary knowledge, offering researchers a reproducible framework to navigate the biases of digital media and utilise cutting-edge computational methods in the study of contentious politics.