Latent Topic Synthesis: Leveraging LLMs for Electoral Ad Analysis

2025-10-16

AI Summary

Political advertising on Meta prior to the 2024 U.S. presidential election poses challenges for scalable, interpretable analysis due to its scale, lack of annotations, and absence of domain-specific seed terms or expert input. Method: We propose an end-to-end unsupervised framework integrating clustering with large language model (LLM) prompting—specifically embedding Moral Foundations Theory into iterative prompt engineering—to automatically generate semantically rich, ethically grounded, and interpretable topic–morality joint labels for unlabeled ads. Contribution/Results: Our approach reveals that vote-mobilization and immigration ads incurred the highest spending, while abortion and election integrity exhibited exceptional reach. Systematic polarization emerged across topics in funding sources, moral framing (e.g., fairness vs. loyalty), and audience-targeting strategies. The framework advances methodology for digital political communication research, offering empirical grounding for policy regulation and media literacy initiatives.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Humans and AI: Voting
• Application Domains: Humanities & Computational Social Science

Application Category

• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Web Mining and Content Analysis: Topic discovery and tracking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Social media platforms play a pivotal role in shaping political discourse, but analyzing their vast and rapidly evolving content remains a major challenge. We introduce an end-to-end framework for automatically generating an interpretable topic taxonomy from an unlabeled corpus. By combining unsupervised clustering with prompt-based labeling, our method leverages large language models (LLMs) to iteratively construct a taxonomy without requiring seed sets or domain expertise. We apply this framework to a large corpus of Meta (previously known as Facebook) political ads from the month ahead of the 2024 U.S. Presidential election. Our approach uncovers latent discourse structures, synthesizes semantically rich topic labels, and annotates topics with moral framing dimensions. We show quantitative and qualitative analyses to demonstrate the effectiveness of our framework. Our findings reveal that voting and immigration ads dominate overall spending and impressions, while abortion and election-integrity achieve disproportionate reach. Funding patterns are equally polarized: economic appeals are driven mainly by conservative PACs, abortion messaging splits between pro- and anti-rights coalitions, and crime-and-justice campaigns are fragmented across local committees. The framing of these appeals also diverges--abortion ads emphasize liberty/oppression rhetoric, while economic messaging blends care/harm, fairness/cheating, and liberty/oppression narratives. Topic salience further reveals strong correlations between moral foundations and issues. Demographic targeting also emerges. This work supports scalable, interpretable analysis of political messaging on social media, enabling researchers, policymakers, and the public to better understand emerging narratives, polarization dynamics, and the moral underpinnings of digital political communication.

Problem

• Automatically generating interpretable topic taxonomies from unlabeled social media content
• Analyzing latent discourse structures and moral framing in political advertising
• Enabling scalable interpretation of polarization dynamics in digital political communication

Innovation

• Combining unsupervised clustering with prompt-based labeling
• Leveraging LLMs to iteratively construct taxonomy
• Automatically generating interpretable topic taxonomy from unlabeled corpus