Mask-to-Correct$^+$: 利用检索器多样性进行掩码引导的忠实事实修正
Mask-to-Correct$^+$: Leveraging Retriever Diversity for Masking-guided Faithful Fact Correction
AI总结：
本研究提出Mask-to-Correct$^+$框架，通过利用检索器多样性来改进掩码引导的事实修正，通过结合多个检索器的修正结果以减少检索偏差并提高鲁棒性，实验表明其在多个基准数据集上均优于现有方法，SARI得分提升达14%。
摘要：
社交媒体上虚假信息的快速传播凸显了需要强大、自动化的事实修正框架。然而，现有方法依赖于监督学习，从人工标注的声明-证据对中学习，这些数据稀缺且易受偏见影响，限制了其在不同领域的泛化能力。此外，这些方法在修正过程中忽略了语义忠实性。为了解决这些挑战，我们提出了Mask-to-Correct (M$_2$C)，一种无需训练、仅在推理时使用的检索增强生成（RAG）基于框架，利用多样性感知掩码来识别声明中的错误片段，并使用检索到的证据评估修正的忠实性。然而，RAG的有效性严重依赖于检索器的选择，这可能因查询而异。为缓解这一问题，我们进一步引入M$_2$C$^+$，一种基于集成的框架，通过结合多个排序器的修正结果以减少检索偏差并提高鲁棒性。在基准数据集上的广泛实验表明，我们提出的框架在不使用黄金证据的情况下，始终优于所有基线方法，SARI得分提升达14%。
英文摘要：
The rapid spread of misinformation on social media highlights the need for robust, automated fact correction frameworks. However, existing works rely on supervised learning from manually annotated claim-evidence pairs, which are scarce and prone to biases, limiting their generalization across domains. Moreover, these methods overlook semantic faithfulness in their correction process. To address these challenges, we propose Mask-to-Correct (M$_2$C), a training-free, inference-only Retrieval Augmented Generation (RAG) based framework that leverages diversity-aware masking to identify erroneous spans of claims and evaluate the faithfulness of corrections using retrieved evidence. However, the effectiveness of RAG heavily depends on the choice of retriever, which may vary across queries. To mitigate this, we further introduce M$_2$C$^+$, an ensemble-based framework that combines corrections across multiple rankers to reduce retrieval bias and improve robustness. Extensive experiments on the benchmark datasets demonstrate that our proposed frameworks consistently outperform all baselines, achieving up to 14% improvement in SARI scores, without using gold evidence.
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。
