Analysing Personal Attacks in U.S. Presidential Debates

2025-11-14

AI Summary

This study addresses the automatic identification of personal attacks in U.S. presidential debates. Methodologically, it introduces the first cross-election framework for systematic annotation and analysis, built upon a high-quality manually annotated corpus; task-specific adaptation strategies are applied to fine-tune Transformer models (e.g., BERT), while prompting-based fine-grained detection capabilities of large language models (LLMs) are systematically explored. Key contributions include: (1) the first structured annotation of implicit aggressive utterances across multiple presidential debate cycles; (2) a novel language model optimization approach tailored to formal political discourse; and (3) substantial improvements in detecting indirect and rhetorical attacks. Experiments demonstrate strong robustness and interpretability on political texts, offering a reusable technical toolkit and empirical foundation for media monitoring, public perception research, and political communication studies.

Technology Category

• Natural Language Processing: Discourse, Pragmatics & Argument Mining
• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Personal attacks have become a notable feature of U.S. presidential debates and play an important role in shaping public perception during elections. Detecting such attacks can improve transparency in political discourse and provide insights for journalists, analysts and the public. Advances in deep learning and transformer-based models, particularly BERT and large language models (LLMs) have created new opportunities for automated detection of harmful language. Motivated by these developments, we present a framework for analysing personal attacks in U.S. presidential debates. Our work involves manual annotation of debate transcripts across the 2016, 2020 and 2024 election cycles, followed by statistical and language-model based analysis. We investigate the potential of fine-tuned transformer models alongside general-purpose LLMs to detect personal attacks in formal political speech. This study demonstrates how task-specific adaptation of modern language models can contribute to a deeper understanding of political communication.

Problem

• Detecting personal attacks in presidential debates
• Analyzing harmful language using transformer models
• Improving political discourse transparency through AI

Innovation

• Fine-tuned transformer models detect personal attacks
• Manual annotation of presidential debate transcripts
• Statistical and language-model based analysis framework