@inproceedings{loftus-etal-2025-using,
title = "Using {LLM}s and Preference Optimization for Agreement-Aware {H}ate{W}i{C} Classification",
author = {Loftus, Sebastian and
M{\"u}lthaler, Adrian and
Hoeken, Sanne and
Zarrie{\ss}, Sina and
Alacam, Ozge},
editor = "Calabrese, Agostina and
de Kock, Christine and
Nozza, Debora and
Plaza-del-Arco, Flor Miriam and
Talat, Zeerak and
Vargas, Francielle",
booktitle = "Proceedings of the The 9th Workshop on Online Abuse and Harms (WOAH)",
month = aug,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.woah-1.47/",
pages = "538--547",
ISBN = "979-8-89176-105-6",
abstract = "Annotator disagreement poses a significant challenge in subjective tasks like hate speech detection. In this paper, we introduce a novel variant of the HateWiC task that explicitly models annotator agreement by estimating the proportion of annotators who classify the meaning of a term as hateful. To tackle this challenge, we explore the use of Llama 3 models fine-tuned through Direct Preference Optimization (DPO). Our experiments show that while LLMs perform well for majority-based hate classification, they struggle with the more complex agreement-aware task. DPO fine-tuning offers improvements, particularly when applied to instruction-tuned models. Yet, our results emphasize the need for improved modeling of subjectivity in hate classification and this study can serve as foundation for future advancements."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="loftus-etal-2025-using">
<titleInfo>
<title>Using LLMs and Preference Optimization for Agreement-Aware HateWiC Classification</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sebastian</namePart>
<namePart type="family">Loftus</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Adrian</namePart>
<namePart type="family">Mülthaler</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sanne</namePart>
<namePart type="family">Hoeken</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sina</namePart>
<namePart type="family">Zarrieß</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ozge</namePart>
<namePart type="family">Alacam</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the The 9th Workshop on Online Abuse and Harms (WOAH)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Agostina</namePart>
<namePart type="family">Calabrese</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christine</namePart>
<namePart type="family">de Kock</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Debora</namePart>
<namePart type="family">Nozza</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Flor</namePart>
<namePart type="given">Miriam</namePart>
<namePart type="family">Plaza-del-Arco</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zeerak</namePart>
<namePart type="family">Talat</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Francielle</namePart>
<namePart type="family">Vargas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Vienna, Austria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-105-6</identifier>
</relatedItem>
<abstract>Annotator disagreement poses a significant challenge in subjective tasks like hate speech detection. In this paper, we introduce a novel variant of the HateWiC task that explicitly models annotator agreement by estimating the proportion of annotators who classify the meaning of a term as hateful. To tackle this challenge, we explore the use of Llama 3 models fine-tuned through Direct Preference Optimization (DPO). Our experiments show that while LLMs perform well for majority-based hate classification, they struggle with the more complex agreement-aware task. DPO fine-tuning offers improvements, particularly when applied to instruction-tuned models. Yet, our results emphasize the need for improved modeling of subjectivity in hate classification and this study can serve as foundation for future advancements.</abstract>
<identifier type="citekey">loftus-etal-2025-using</identifier>
<location>
<url>https://aclanthology.org/2025.woah-1.47/</url>
</location>
<part>
<date>2025-08</date>
<extent unit="page">
<start>538</start>
<end>547</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Using LLMs and Preference Optimization for Agreement-Aware HateWiC Classification
%A Loftus, Sebastian
%A Mülthaler, Adrian
%A Hoeken, Sanne
%A Zarrieß, Sina
%A Alacam, Ozge
%Y Calabrese, Agostina
%Y de Kock, Christine
%Y Nozza, Debora
%Y Plaza-del-Arco, Flor Miriam
%Y Talat, Zeerak
%Y Vargas, Francielle
%S Proceedings of the The 9th Workshop on Online Abuse and Harms (WOAH)
%D 2025
%8 August
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-105-6
%F loftus-etal-2025-using
%X Annotator disagreement poses a significant challenge in subjective tasks like hate speech detection. In this paper, we introduce a novel variant of the HateWiC task that explicitly models annotator agreement by estimating the proportion of annotators who classify the meaning of a term as hateful. To tackle this challenge, we explore the use of Llama 3 models fine-tuned through Direct Preference Optimization (DPO). Our experiments show that while LLMs perform well for majority-based hate classification, they struggle with the more complex agreement-aware task. DPO fine-tuning offers improvements, particularly when applied to instruction-tuned models. Yet, our results emphasize the need for improved modeling of subjectivity in hate classification and this study can serve as foundation for future advancements.
%U https://aclanthology.org/2025.woah-1.47/
%P 538-547
Markdown (Informal)
[Using LLMs and Preference Optimization for Agreement-Aware HateWiC Classification](https://aclanthology.org/2025.woah-1.47/) (Loftus et al., WOAH 2025)
ACL