大语言模型中常识推理的跨文化迁移：来自阿拉伯世界的证据
Cross-Cultural Transfer of Commonsense Reasoning in LLMs: Evidence from the Arab World
AI总结：
本文研究大语言模型常识推理的跨文化迁移，通过在13个阿拉伯国家数据集上评估轻量级对齐方法，发现仅用12个文化特定示例即可平均提升10%性能，且非本文化示范可达到或超过本文化对齐效果，证明高效跨文化对齐可行。
摘要：
大语言模型（LLMs）常常反映以西方为中心的偏见，限制了它们在多元文化语境中的有效性。尽管已有一些工作探讨了文化对齐，但利用一种文化中的对齐来改善其他文化中表现的跨文化迁移潜力仍未得到充分探索。本文研究了阿拉伯世界中常识推理的跨文化迁移，该地区语言和历史相似性与地方文化差异并存。使用一个覆盖13个阿拉伯国家的基于文化语境的常识推理数据集，我们评估了轻量级对齐方法，如上下文学习（in-context learning）和基于示范的强化学习（DITTO），并与监督微调（supervised fine-tuning）和直接偏好优化（direct preference optimization）等基线方法进行比较。我们的结果表明，在多语言模型中，仅来自一个国家12个文化特定示例就能使其他国家平均性能提升10%。此外，我们证明来自印度尼西亚和美国语境的非本文化示范在多项选择题（MCQ）推理上可以达到或超过本文化对齐效果，凸显了文化常识在阿拉伯世界之外的可迁移性。这些发现表明高效的跨文化对齐是可行的，并为将大语言模型适配到低资源文化环境提供了一种有前景的方法。
英文摘要：
Large language models (LLMs) often reflect Western-centric biases, limiting their effectiveness in diverse cultural contexts. Although some work has explored cultural alignment, the potential for cross-cultural transfer, using alignment in one culture to improve performance in others, remains underexplored. This paper investigates cross-cultural transfer of commonsense reasoning in the Arab world, where linguistic and historical similarities coexist with local cultural differences. Using a culturally grounded commonsense reasoning dataset covering 13 Arab countries, we evaluate lightweight alignment methods such as in-context learning and demonstration-based reinforcement (DITTO), alongside baselines like supervised fine-tuning and direct preference optimization. Our results show that merely 12 culture-specific examples from one country can improve performance in others by 10\% on average, within multilingual models. In addition, we demonstrate that out-of-culture demonstrations from Indonesia and US contexts can match or surpass in-culture alignment for MCQ reasoning, highlighting cultural commonsense transferability beyond the Arab world. These findings demonstrate that efficient cross-cultural alignment is possible and offer a promising approach to adapt LLMs to low-resource cultural settings.
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。
