让模型从原始文本中自动识别零指代，实现端到端共指消解。
CorPipe at CRAC 2024: Predicting Zero Mentions from Raw Text
- 分两阶段预测空节点，再联合处理共指提及与链接。
- 单阶段模型直接联合生成空节点与共指关系，准确率领先2.8个百分点。
- 适用于需要处理无标注空指代的多语言文本理解任务。
我们介绍 CorPipe 24，这是在 CRAC 2024 多语言共指消解共享任务中的获胜方案。本届任务首次要求在原始文本上预测零指代所需的空节点（往年任务中空节点作为输入给出）。为此，我们评估了两种模型变体：两阶段方法（先用预训练编码器预测空节点，再由另一预训练模型联合处理句子词与空节点）和单阶段方法（单一预训练编码器联合生成空节点、共指提及及共指链接）。在两种设置下，CorPipe 分别以 3.9 和 2.8 个百分点的优势超越其他参赛者。源代码与训练模型已开源于 https://github.com/ufal/crac2024-corpipe。
原文摘要 · Abstract (English)
We present CorPipe 24, the winning entry to the CRAC 2024 Shared Task on Multilingual Coreference Resolution. In this third iteration of the shared task, a novel objective is to also predict empty nodes needed for zero coreference mentions (while the empty nodes were given on input in previous years). This way, coreference resolution can be performed on raw text. We evaluate two model variants: a~two-stage approach (where the empty nodes are predicted first using a pretrained encoder model and then processed together with sentence words by another pretrained model) and a single-stage approach (where a single pretrained encoder model generates empty nodes, coreference mentions, and coreference links jointly). In both settings, CorPipe surpasses other participants by a large margin of 3.9 and 2.8 percent points, respectively. The source code and the trained model are available at https://github.com/ufal/crac2024-corpipe.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。