一种基于跨距的构成树线性化方法
A Span-based Linearization for Constituent Trees
摘要：
我们提出了一种新的线性化的组成树，连同一个新的局部归一化模型。对于句子中的每个分割点，我们的模型计算以该分割点结束的所有跨度上的规范化子，然后从它们预测树跨度。与全局模型相比，我们的模型是快速和并行化。与以往的局部模型不同，本文的线性化方法直接与跨度相联系，在进行跨度预测时考虑了更多的局部特征，更具有可解释性和有效性。在PTB（95.8 F1）和CTB（92.4 F1）上的实验表明，我们的模型显著优于现有的局部模型，并有效地实现了与全局模型竞争的结果。
英文摘要：
We propose a novel linearization of a constituent tree, together with a new locally normalized model. For each split point in a sentence, our model computes the normalizer on all spans ending with that split point, and then predicts a tree span from them. Compared with global models, our model is fast and parallelizable. Different from previous local models, our linearization method is tied on the spans directly and considers more local features when performing span prediction, which is more interpretable and effective. Experiments on PTB (95.8 F1) and CTB (92.4 F1) show that our model significantly outperforms existing local models and efficiently achieves competitive results with global models.
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。
