Cross-Lingual Dependency Parsing by POS-Guided Word Reordering

Findings of the Association for Computational Linguistics: EMNLP 2020, pages 2938–2948 November 16 - 20, 2020. c⃝2020 Association for Computational Linguistics 2938 Cross-Lingual Dependency Parsing by POS-Guided Word Reordering Lu Liu1,2, Yi Zhou1,2, Jianhan Xu1,2, Xiaoqing Zheng1,2, Kai-Wei Chang3, Xuanjing Huang1,2 1School of Computer Science, Fudan University, Shanghai, China 2Shanghai Key Laboratory of Intelligent Information Processing 3University of California, Los Angeles, USA {luliu19, yizhou17, xujh16, zhengxq}@fudan.edu.cn kwchang@cs.ucla.edu, xjhuang@fudan.edu.cn Abstract We propose a novel approach to cross-lingual dependency parsing based on word reorder- ing. The words in each sentence of a source language corpus are rearranged to meet the word order in a target language under the guid- ance of a part-of-speech based language model (LM). To obtain the highest reordering score under the LM, a population-based optimiza- tion algorithm and its genetic operators are de- signed to deal with the combinatorial nature of such word reordering. A parser trained on the reordered corpus then can be used to parse sentences in the target language. We demon- strate through extensive experimentation that our approach achieves better or comparable results across 25 target languages (1.73% in- crease in average), and outperforms a baseline by a significant margin on the languages that are greatly different from the source one. For example, when transferring the English parser to Hindi and Latin, our approach outperforms the baseline by 15.3% and 6.7% respectively. 1 Introduction The rise of machine learning (ML) methods and the availability of treebanks (Buchholz and Marsi, 2006) for a wide variety of languages have led to a rapid increase in research on data-driven depen- dency parsing (McDonald and Pereira, 2006; Nivre, 2008; Kiperwasser and Goldberg, 2016). However, the performance of dependency parsers heavily re- lies on the size of corpus. Due to the great cost and difficulty of acquiring sufficient training data, ML-based methods cannot be trivially applied to low-resource languages. Cross-lingual transfer is a promising approach to tackle the lack of sufficient data. The idea is to train a cross-lingual model that transfers knowledge learned in one or multiple high-resource source lan- guages to target ones. This approach has been suc- cessfully applied in various tasks, including part- of-speech (POS) tagging (Kim et al., 2017), depen- dency parsing (McDonald et al., 2011), named en- tity recognition (Xie et al., 2018), entity linking (Sil et al., 2018), question answering (Joty et al., 2017), and coreference resolution (Kundu et al., 2018). A key challenge for cross-lingual parsing is the difficulty to handle word order difference between source and target languages, which often causes a significant drop in performance (Rasooli and Collins, 2017; Ahmad et al., 2019). Inspired by the idea that POS sequences often reflect the syntac- tic structure of a language, we propose CURSOR (Cross lingUal paRSing by wOrd Reordering) to overcome the word order difference issue in cross- lingual transfer. Specifically, we assume we have a treebank in the source language and annotated POS corpus in the target language1. We first train a POS-based language model on a corpus in the …

Content truncated to 500 words