arrow
返回

Nucleus Composition in Transition-based Dependency Parsing

delete2022-12-01
delete1
delete
OA
AI
J
Joakim Nivre *
A
Ali Basirat
L
Luise Dürlich
A
A. Moss
DOI:10.1162/coli_a_00450delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
Dependency-based approaches to syntactic analysis assume that syntactic structure can be analyzed in terms of binary asymmetric dependency relations holding between elementary syntactic units. Computational models for dependency parsing almost universally assume that an elementary syntactic unit is a word, while the influential theory of Lucien Tesniere instead posits a more abstract notion of nucleus, which may be realized as one or more words. In this article, we investigate the effect of enriching computational parsing models with a concept of nucleus inspired by Tesniere. We begin by reviewing how the concept of nucleus can be defined in the framework of Universal Dependencies, which has become the de facto standard for training and evaluating supervised dependency parsers, and explaining how composition functions can be used to make neural transition-based dependency parsers aware of the nuclei thus defined. We then perform an extensive experimental study, using data from 20 languages to assess the impact of nucleus composition across languages with different typological characteristics, and utilizing a variety of analytical tools including ablation, linear mixed-effects models, diagnostic classifiers, and dimensionality reduction. The analysis reveals that nucleus composition gives small but consistent improvements in parsing accuracy for most languages, and that the improvement mainly concerns the analysis of main predicates, nominal dependents, clausal dependents, and coordination structures. Significant factors explaining the rate of improvement across languages include entropy in coordination structures and frequency of certain function words, in particular determiners. Analysis using dimensionality reduction and diagnostic classifiers suggests that nucleus composition increases the similarity of vectors representing nuclei of the same syntactic type.

期刊

Computational Linguistics 封面图
Computational Linguistics
IF:
5.3
论文数:
837
被引数:
2.7K

机构

L
Linkoping University
学者数:
1.6W
论文数: 1.5W
被引数: 184
R
RISE Research Institutes of Sweden
学者数:
1.9K
论文数: 1.5K
被引数: 2.5K
U
uppsala university
学者数:
3.7W
论文数: 3.4W
被引数: 47
学者 查看更多机构
引用论文

引用论文

New aspects of infantile oxalosis
err1987-01-01
err0
PREAI
errErnst P. Leumann; Alois Niederwieser; Andreas Fanconi
err分享
err收藏
Summary of the Meetings
err2007-12-28
err0
PREAI
errMaureen A. Donnelly
err分享
err收藏
err分享
err收藏
Fitting Linear Mixed-Effects Models Using lme4使用lme4拟合线性混合效应模型
err2015-01-01
err6.0W
errOAAI
errBates, Douglas; Maechler, Martin; Bolker, Benjamin M.; Walker, Steven C.
err分享
err收藏
Multiple Alpha Chain Loci for Human Haemoglobins: Hb J-Buda and Hb G-Pest
err1972-01-01
err0
PREAI
errS. R. HOLLÁN; J. G. SZELENYI; B. BRIMHALL; M. DUERST; R. T. JONES; R. D. KOLER; Z. STOCKLEN
err分享
err收藏
Quantitative trait loci for sugarcane resistance to the spotted stem borer Chilo sacchariphagus
err2010-11-18
err0
PREAI
errS. Nibouche; L. M. Raboin; J.-Y. Hoarau; A. D’Hont; L. Costet
err分享
err收藏
err分享
err收藏
学者 查看更多内容