A Few Hypocrites: Few-Shot Learning and Subtype Definitions for Detecting Hypocrisy Accusations in Online Climate Change Debates
This work addresses the need for large-scale text analysis tools in climate change discourse research, though it is incremental as it builds on existing fallacious argument detection methods.
The paper tackles the problem of detecting hypocrisy accusations in online climate change debates by defining it as an independent NLP task and creating a corpus of 420 Reddit comments annotated for personal and political hypocrisy subtypes. Results show that few-shot learning with LLMs like GPT-4o and Llama-3 achieves an F1 score of 0.68, outperforming previous work with 0.44, though models struggle more with political hypocrisy.
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.