A Few Hypocrites: Few-Shot Learning and Subtype Definitions for Detecting Hypocrisy Accusations in Online Climate Change Debates

TL;DR

This paper introduces a new NLP task for detecting hypocrisy accusations in online climate debates, creating a specialized dataset and evaluating large language models, which show promising results but struggle with complex political accusations.

Contribution

It defines hypocrisy accusation detection as an independent task, develops the Climate Hypocrisy Accusation Corpus, and evaluates few-shot learning with large language models for this purpose.

Findings

GPT-4o and Llama-3 achieve F1 scores up to 0.68.

Models perform better on personal hypocrisy than political.

Context is crucial for accurately detecting hypocrisy accusations.

Abstract

The climate crisis is a salient issue in online discussions, and hypocrisy accusations are a central rhetorical element in these debates. However, for large-scale text analysis, hypocrisy accusation detection is an understudied tool, most often defined as a smaller subtask of fallacious argument detection. In this paper, we define hypocrisy accusation detection as an independent task in NLP, and identify different relevant subtypes of hypocrisy accusations. Our Climate Hypocrisy Accusation Corpus (CHAC) consists of 420 Reddit climate debate comments, expert-annotated into two different types of hypocrisy accusations: personal versus political hypocrisy. We evaluate few-shot in-context learning with 6 shots and 3 instruction-tuned Large Language Models (LLMs) for detecting hypocrisy accusations in this dataset. Results indicate that the GPT-4o and Llama-3 models in particular show promise in detecting hypocrisy accusations (F1 reaching 0.68, while previous work shows F1 of 0.44). However, context matters for a complex semantic concept such as hypocrisy accusations, and we find models struggle especially at identifying political hypocrisy accusations compared to personal moral hypocrisy. Our study contributes new insights in hypocrisy detection and climate change discourse, and is a stepping stone for large-scale analysis of hypocrisy accusation in online climate debates.