A Dynamic Oracle for Arc-Eager Dependency Parsing
作者： Joakim Nivre , Yoav Goldberg
DOI:
关键词: Parsing 、 Algorithm 、 Oracle 、 Parser combinator 、 Sentence 、 LR parser 、 Tree (data structure) 、 Computer science 、 Set (abstract data type) 、 Theoretical computer science 、 Dependency grammar 、 Computational linguistics 、 Transition system
摘要: The standard training regime for transition-based dependency parsers makes use of an oracle, which predicts optimal transition sequence a sentence and its gold tree. We present improved oracle the arc-eager system, provides set transitions every valid parser configuration, including configurations from tree is not reachable. In such cases, that will lead to best reachable given configuration. efficient implement provably correct. train deterministic left-to-right less sensitive error propagation, using online procedure also explores resulting non-optimal sequences transitions. This new outperforms greedy trained conventional oracles on range data sets, with average improvement over 1.2 LAS points up almost 3 some sets.
aclweb.org参考文章(42)
Bernd Bohnet, Jonas Kuhn, The Best of BothWorlds -- A Graph-based Completion Model for Transition-based Parsers conference of the european chapter of the association for computational linguistics. pp. 77- 87 ,(2012)
Vitor R. Carvalho, William W. Cohen, Stacked sequential learning international joint conference on artificial intelligence. pp. 671- 676 ,(2005)
Kemal Oflazer, Bilge Say, Dilek Zeynep Hakkani-Tür, Gökhan Tür, Building a Turkish Treebank Treebanks. pp. 261- 277 ,(2003) , 10.1007/978-94-010-0201-1_15
Joakim Nivre, An efficient algorithm for projective dependency parsing international workshop/conference on parsing technologies. pp. 149- 160 ,(2003)
Alena Böhmová, Jan Hajič, Eva Hajičová, Barbora Hladká, The Prague Dependency Treebank Treebanks. pp. 103- 127 ,(2003) , 10.1007/978-94-010-0201-1_7
Michael A. Covington, A Fundamental Algorithm for Dependency Parsing ,(2004)
Liang Huang, Suphan Fayong, Yang Guo, Structured Perceptron with Inexact Search north american chapter of the association for computational linguistics. pp. 142- 151 ,(2012)
Yuji Matsumoto, Hiroyasu Yamada, Statistical Dependency Analysis with Support Vector Machines international workshop/conference on parsing technologies. pp. 195- 206 ,(2003)
Joakim Nivre, Ryan McDonald, Characterizing the Errors of Data-Driven Dependency Parsing Models empirical methods in natural language processing. pp. 122- 131 ,(2007)
Joakim Nivre, Johan Hall, Jens Nilsson, Memory-Based Dependency Parsing conference on computational natural language learning. pp. 49- 56 ,(2004)