MFTCXplain: A Multilingual Benchmark Dataset for Evaluating the Moral Reasoning of LLMs through Multi-hop Hate Speech Explanation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Trager, Jackson, Vargas, Francielle, Alves, Diego, Guida, Matteo, Ngueajio, Mikel K., Agrawal, Ameeta, Daryani, Yalda, Karimi-Malekabadi, Farzan, Plaza-del-Arco, Flor Miriam
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915547917058048
author Trager, Jackson
Vargas, Francielle
Alves, Diego
Guida, Matteo
Ngueajio, Mikel K.
Agrawal, Ameeta
Daryani, Yalda
Karimi-Malekabadi, Farzan
Plaza-del-Arco, Flor Miriam
author_facet Trager, Jackson
Vargas, Francielle
Alves, Diego
Guida, Matteo
Ngueajio, Mikel K.
Agrawal, Ameeta
Daryani, Yalda
Karimi-Malekabadi, Farzan
Plaza-del-Arco, Flor Miriam
contents Ensuring the moral reasoning capabilities of Large Language Models (LLMs) is a growing concern as these systems are used in socially sensitive tasks. Nevertheless, current evaluation benchmarks present two major shortcomings: a lack of annotations that justify moral classifications, which limits transparency and interpretability; and a predominant focus on English, which constrains the assessment of moral reasoning across diverse cultural settings. In this paper, we introduce MFTCXplain, a multilingual benchmark dataset for evaluating the moral reasoning of LLMs via multi-hop hate speech explanation using the Moral Foundations Theory. MFTCXplain comprises 3,000 tweets across Portuguese, Italian, Persian, and English, annotated with binary hate speech labels, moral categories, and text span-level rationales. Our results show a misalignment between LLM outputs and human annotations in moral reasoning tasks. While LLMs perform well in hate speech detection (F1 up to 0.836), their ability to predict moral sentiments is notably weak (F1 < 0.35). Furthermore, rationale alignment remains limited mainly in underrepresented languages. Our findings show the limited capacity of current LLMs to internalize and reflect human moral reasoning
format Preprint
id arxiv_https___arxiv_org_abs_2506_19073
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MFTCXplain: A Multilingual Benchmark Dataset for Evaluating the Moral Reasoning of LLMs through Multi-hop Hate Speech Explanation
Trager, Jackson
Vargas, Francielle
Alves, Diego
Guida, Matteo
Ngueajio, Mikel K.
Agrawal, Ameeta
Daryani, Yalda
Karimi-Malekabadi, Farzan
Plaza-del-Arco, Flor Miriam
Computation and Language
Ensuring the moral reasoning capabilities of Large Language Models (LLMs) is a growing concern as these systems are used in socially sensitive tasks. Nevertheless, current evaluation benchmarks present two major shortcomings: a lack of annotations that justify moral classifications, which limits transparency and interpretability; and a predominant focus on English, which constrains the assessment of moral reasoning across diverse cultural settings. In this paper, we introduce MFTCXplain, a multilingual benchmark dataset for evaluating the moral reasoning of LLMs via multi-hop hate speech explanation using the Moral Foundations Theory. MFTCXplain comprises 3,000 tweets across Portuguese, Italian, Persian, and English, annotated with binary hate speech labels, moral categories, and text span-level rationales. Our results show a misalignment between LLM outputs and human annotations in moral reasoning tasks. While LLMs perform well in hate speech detection (F1 up to 0.836), their ability to predict moral sentiments is notably weak (F1 < 0.35). Furthermore, rationale alignment remains limited mainly in underrepresented languages. Our findings show the limited capacity of current LLMs to internalize and reflect human moral reasoning
title MFTCXplain: A Multilingual Benchmark Dataset for Evaluating the Moral Reasoning of LLMs through Multi-hop Hate Speech Explanation
topic Computation and Language
url https://arxiv.org/abs/2506.19073