@inproceedings{weihua-etal-2025-ccl,
    title = "{CCL}-{XC}o{T}: An Efficient Cross-Lingual Knowledge Transfer Method for Mitigating Hallucination Generation",
    author = "Weihua, Zheng  and
      Lee, Roy Ka-Wei  and
      Liu, Zhengyuan  and
      Kui, Wu  and
      Aw, AiTi  and
      Zou, Bowei",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/author-page-yu-wang-polytechnic/2025.findings-emnlp.93/",
    doi = "10.18653/v1/2025.findings-emnlp.93",
    pages = "1768--1788",
    ISBN = "979-8-89176-335-7",
    abstract = "Multilingual Large Language Models (MLLMs) demonstrate strong generalization across languages, yet they remain prone to hallucinations, especially in low-resource languages, due to training data imbalances. These hallucinations, which include inaccurate or fabricated outputs, are particularly problematic in domain-specific generation tasks (Chataigner et al., 2024). To address this challenge, we propose CCL-XCoT (Curriculum-based Contrastive Learning-based Cross-lingual Chain-of-Thought), a two-stage fine-tuning framework for mitigating hallucination in MLLMs. Our approach first enhances cross-lingual semantic alignment through curriculum-based contrastive learning combined with next-token prediction during continued pre-training. Building on this foundation, we then introduce a cross-lingual Chain-of-Thought (XCoT) prompting strategy during instruction fine-tuning, which guides the model to reason in a high-resource language before generating answers in the target low-resource language. Experimental results show that CCL-XCoT reduces hallucination rates by up to 62{\%} and substantially improves factual knowledge transfer across language pairs, without relying on external retrieval or multi-model ensembles."
}