Is LLM an Overconfident Judge? Unveiling the Capabilities of LLMs in Detecting Offensive Language with Annotation Disagreement

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lu, Junyu, Ma, Kai, Wang, Kaichun, Xiao, Kelaiti, Lee, Roy Ka-Wei, Xu, Bo, Yang, Liang, Lin, Hongfei
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916741316083712
author Lu, Junyu
Ma, Kai
Wang, Kaichun
Xiao, Kelaiti
Lee, Roy Ka-Wei
Xu, Bo
Yang, Liang
Lin, Hongfei
author_facet Lu, Junyu
Ma, Kai
Wang, Kaichun
Xiao, Kelaiti
Lee, Roy Ka-Wei
Xu, Bo
Yang, Liang
Lin, Hongfei
contents Large Language Models (LLMs) have become essential for offensive language detection, yet their ability to handle annotation disagreement remains underexplored. Disagreement samples, which arise from subjective interpretations, pose a unique challenge due to their ambiguous nature. Understanding how LLMs process these cases, particularly their confidence levels, can offer insight into their alignment with human annotators. This study systematically evaluates the performance of multiple LLMs in detecting offensive language at varying levels of annotation agreement. We analyze binary classification accuracy, examine the relationship between model confidence and human disagreement, and explore how disagreement samples influence model decision-making during few-shot learning and instruction fine-tuning. Our findings reveal that LLMs struggle with low-agreement samples, often exhibiting overconfidence in these ambiguous cases. However, utilizing disagreement samples in training improves both detection accuracy and model alignment with human judgment. These insights provide a foundation for enhancing LLM-based offensive language detection in real-world moderation tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2502_06207
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Is LLM an Overconfident Judge? Unveiling the Capabilities of LLMs in Detecting Offensive Language with Annotation Disagreement
Lu, Junyu
Ma, Kai
Wang, Kaichun
Xiao, Kelaiti
Lee, Roy Ka-Wei
Xu, Bo
Yang, Liang
Lin, Hongfei
Computation and Language
Artificial Intelligence
Large Language Models (LLMs) have become essential for offensive language detection, yet their ability to handle annotation disagreement remains underexplored. Disagreement samples, which arise from subjective interpretations, pose a unique challenge due to their ambiguous nature. Understanding how LLMs process these cases, particularly their confidence levels, can offer insight into their alignment with human annotators. This study systematically evaluates the performance of multiple LLMs in detecting offensive language at varying levels of annotation agreement. We analyze binary classification accuracy, examine the relationship between model confidence and human disagreement, and explore how disagreement samples influence model decision-making during few-shot learning and instruction fine-tuning. Our findings reveal that LLMs struggle with low-agreement samples, often exhibiting overconfidence in these ambiguous cases. However, utilizing disagreement samples in training improves both detection accuracy and model alignment with human judgment. These insights provide a foundation for enhancing LLM-based offensive language detection in real-world moderation tasks.
title Is LLM an Overconfident Judge? Unveiling the Capabilities of LLMs in Detecting Offensive Language with Annotation Disagreement
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2502.06207