Modelling Protest Related Topics by Combining GPT-4 with State-of-the-Art Approaches

Abstract

This study explores various topic modelling techniques to analyze social media data and gain insights into a Black Lives Matter demonstration in Amsterdam during the COVID-19 pandemic. Models such as Latent Dirichlet Allocation (LDA) and Hierarchical Dirichlet Process (HDP), enriched with synonyms from the Dutch WordNet and filtered hashtags are used for this purpose. The resulting topics were fed into GPT-4 to generate both general and time-specific descriptions of the events, revealing a strong alignment between the shifts in public sentiment and key events. However, some nuanced details were missed. This research highlights the potential of combining advanced topic modelling techniques with GPT-4 for real-time event monitoring, but it also underscores the challenges of synonym-based analysis.