LLMs and clustering create topic taxonomies from unlabeled text

Researchers have developed a new framework that uses large language models (LLMs) and clustering techniques to automatically create interpretable topic taxonomies from unlabeled text data. This method, demonstrated in a case study analyzing political advertising for the 2024 U.S. presidential election, can organize large text corpora without needing pre-defined labels. The induced taxonomy supports downstream analysis of issue prevalence, moral framing, advertising spend, and demographic exposure. AI

IMPACT Provides a scalable method for analyzing large volumes of unstructured text data, potentially aiding research in political science and social media analysis.

RANK_REASON Academic paper detailing a new methodology for topic modeling using LLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Electoral Advertising
• Large Language Models
• Tunazzina Islam

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →