arXivDaily arXiv每日学术速递 周一至周五更新
​通知:联邦节假日调休:哥伦布日 / 原住民日(Columbus Day / Indigenous Peoples' Day),今日(13日)无更新。
arXiv 2503.02442cs.CL

AILS-NTUA 在 SemEval-2025 任务 3 中的表现:利用大语言模型与翻译策略实现多语言幻觉检测

AILS-NTUA at SemEval-2025 Task 3: Leveraging Large Language Models and Translation Strategies for Multilingual Hallucination Detection

School of Electrical and Computer Engineering(电气与计算机工程学院);AILS Laboratory(AILS实验室);National Technical University of Athens(雅典国家技术大学)

Dimitra Karkani, Maria Lymperaiou, Giorgos Filandrianos, Nikolaos Spanos, Athanasios Voulodimos, Giorgos Stamou

AI总结:

针对多语言幻觉检测探索不足的问题,提出无需训练的LLM提示策略,通过将多语言文本译为英文提升检测效果,在多语言上取得竞争力排名,低资源语言获两项第一,验证了该翻译策略的通用性。

摘要:

多语言幻觉检测是一项尚未被充分探索的挑战,Mu-SHROOM 共享任务正致力于解决这一问题。本文提出了一种高效的、无需训练的 LLM 提示策略,通过将多语言文本片段翻译成英文来提升检测效果。该方法在多种语言上取得了具有竞争力的排名,在低资源语言中斩获两个第一名。结果的一致性凸显了翻译策略在幻觉检测中的有效性,证明其不受源语言限制的适用性。

英文摘要:

Multilingual hallucination detection stands as an underexplored challenge, which the Mu-SHROOM shared task seeks to address. In this work, we propose an efficient, training-free LLM prompting strategy that enhances detection by translating multilingual text spans into English. Our approach achieves competitive rankings across multiple languages, securing two first positions in low-resource languages. The consistency of our results highlights the effectiveness of our translation strategy for hallucination detection, demonstrating its applicability regardless of the source language.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑