This research introduces an innovative methodology to extract latent arguments from social media discussions on contentious topics, overcoming limitations of traditional supervised and unsupervised methods.

Key findings:

• LLMs-in-the-loop approach effectively identifies nuanced arguments that traditional methods miss
• Successfully analyzes dynamic social media conversations despite shifting focus and themes
• Reveals demographic targeting patterns in public discourse
• Provides deeper insights than traditional topic modeling approaches

Security Implications: The methodology enables more effective monitoring of public discourse, helping identify misinformation campaigns and potentially problematic content across social media platforms.

Uncovering Latent Arguments in Social Media Messaging by Employing LLMs-in-the-Loop Strategy