Dependency Parsing as MRC-based Span-Span Prediction

Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 2427 - 2437 May 22-27, 2022 c⃝2022 Association for Computational Linguistics Dependency Parsing as MRC-based Span-Span Prediction Leilei Gan♦, Yuxian Meng♣, Kun Kuang♦, Xiaofei Sun♦♣ Chun Fan♠, Fei Wu♦n and Jiwei Li♦♣ ♦College of Computer Science and Technology, Zhejiang University, ♣Shannon.AI nShanghai Institute for Advanced Study of Zhejiang University, nShanghai AI Laboratory ♠Peng Cheng Laboratory, ♠National Biomedical Imaging Center, Peking University ♠ Computer Center, Peking University {leileigan, kunkuang, wufei, jiwei_li}@zju.edu.cn {yuxian_meng, xiaofei_sun}@shannonai.com, fanchun@pku.edu.cn Abstract Higher-order methods for dependency parsing can partially but not fully address the issue that edges in dependency trees should be con- structed at the text span/subtree level rather than word level. In this paper, we propose a new method for dependency parsing to address this issue. The proposed method constructs de- pendency trees by directly modeling span-span (in other words, subtree-subtree) relations. It consists of two modules: the text span pro- posal module which proposes candidate text spans, each of which represents a subtree in the dependency tree denoted by (root, start, end); and the span linking module, which con- structs links between proposed spans. We use the machine reading comprehension (MRC) framework as the backbone to formalize the span linking module, where one span is used as query to extract the text span/subtree it should be linked to. The proposed method has the following merits: (1) it addresses the fun- damental problem that edges in a dependency tree should be constructed between subtrees; (2) the MRC framework allows the method to retrieve missing spans in the span proposal stage, which leads to higher recall for eligi- ble spans. Extensive experiments on the PTB, CTB and Universal Dependencies (UD) bench- marks demonstrate the effectiveness of the pro- posed method. 1 2 1 Introduction Dependency parsing is a basic and fundamental task in natural language processing (NLP) (Eis- ner, 2000; Nivre, 2003; McDonald et al., 2005b). Among existing efforts for dependency parsers, graph-based models (McDonald et al., 2005a; Pei et al., 2015) are a widely used category of models, which cast the task as finding the optimal maxi- mum spanning tree in the directed graph. Graph- 1Chun Fan is the corresponding author. 2The code is available at https://github.com/ ShannonAI/mrc-for-dependency-parsing I love Tim’s cat nsubj nmod obj root I love Tim’s cat nsubj obj obj root Figure 1: Two possible dependency trees for sentence “I love Tim’s cat”. For the tree on the right hand side, if we look at the token-token level, ‘love" being linked to “cat" is correct. But at the subtree-subtree level, the linking is incorrect since the span/subtree behind “cat" is incorrect. based models provide a more global view than shift- reduce models (Zhang and Nivre, 2011; Chen and Manning, 2014), leading to better performances. Graph-based methods are faced with a challenge: they construct dependency edges by using word pairs as basic units for modeling, which is insuffi- cient because dependency parsing performs at the span/subtree level. …

Content truncated to 500 words