Skip to content

M
mtbookv2
Commit 83d21178 by 曹润柘
Options

合并分支 'caorunzhe' 到 'master' 

Caorunzhe

查看合并请求 !173
parents 2631bdd6 788309b0
正在显示 包含 470 行增加 和 12 行删除
Chapter8/chapter8.tex
......@@ -1583,19 +1583,21 @@ d_1 = {d'} \circ {r_5}
\sectionnewpage
\section{小结及深入阅读}
\parinterval \color{red}统计机器翻译模型是近三十年内自然语言处理的重要里程碑之一。其统计建模的思想长期影响着自然语言处理的研究。无论是基于短语的模型，还是基于层次短语的模型，还是基于语言学句法的模型都在尝试回答：究竟应该用什么样的知识对机器翻译进行统计建模？不过，这个问题至今还没有确定的答案。但是，显而易见，统计机器翻译为机器翻译的研究提供了一种范式，即让计算机用概率化的“知识”描述翻译问题。这些“ 知识”就是统计模型的参数，模型可以从大量的双语和单语数据中自动学习参数。这种建模思想在今天的机器翻译研究中仍然随处可见。
\parinterval 自基于规则的方法开始，如何句法信息就是机器翻译研究人员关注的热点。在统计机器翻译时代，句法信息与机器翻译的结合成为了最具时态特色的研究方向之一。句法结构具有高度的抽象性，因此可以缓解基于词串方法不善于处理句子上层结构的问题。
\parinterval 本章对统计机器翻译的经典模型进行了介绍。从早期的基于短语的模型，再到层次短语模型，以及更为复杂的基于语言学句法的模型，本章尝试对不同的建模思想进行阐释。只是，统计机器翻译的内容非常丰富，很难通过一章的内容进行面面俱到的介绍。还有很多方向值得读者进一步了解：
\parinterval 本章对基于句法的机器翻译模型进行了介绍，并重点讨论了相关的建模、翻译规则抽取以及解码问题。从某种意义上说，基于句法的模型与基于短语的模型都同属一类模型，因为二者都假设：两种语言间存在由短语或者规则构成的翻译推导，而机器翻译的目标就是找到最优的翻译推导。但是，由于句法信息有其独特的性质，因此也给机器翻译带来了新的问题。有几方面问题值得关注：
\begin{itemize}
\vspace{0.5em}
\item 统计机器翻译的成功很大程度上来自判别式模型引入任意特征的能力。因此，在统计机器翻译时代，很多工作都集中在新特征的设计上。比如，可以基于不同的统计特征和先验知识设计翻译特征\upcite{och2004smorgasbord,Chiang200911,gildea2003loosely}，也可以模仿分类任务设计大规模的稀疏特征\upcite{chiang2008online}。另一方面，模型训练和特征权重调优也是统计机器翻译中的重要问题，除了最小错误率训练，还有很多方法，比如，最大似然估计\upcite{koehn2003statistical,Peter1993The}、判别式方法\upcite{Blunsom2008A}、贝叶斯方法\upcite{Blunsom2009A,Cohn2009A}、最小风险训练\upcite{smith2006minimum,li2009first-}、基于Margin的方法\upcite{watanabe2007online,Chiang200911}以及基于排序模型的方法（PRO）\upcite{Hopkins2011Tuning,dreyer2015apro}。实际上，统计机器翻译的训练和解码也存在不一致的问题，比如，特征值由双语数据上的极大似然估计得到（没有剪枝），而解码时却使用束剪枝，而且模型的目标是最大化机器翻译评价指标。对于这个问题也可以通过调整训练的目标函数进行缓解\upcite{XiaoA,marcu2006practical}。
\item 从建模的角度看，早期的统计机器翻译模型已经涉及到了树结构的表示问题\upcite{DBLP:conf/acl/AlshawiBX97,DBLP:conf/acl/WangW98}。不过，基于句法的翻译模型的真正崛起还源自同步文法的提出。初期的工作大多集中在反向转录文法和括号转录文法方面\upcite{DBLP:conf/acl-vlc/Wu95,wu1997stochastic,DBLP:conf/acl/WuW98}，这类方法也被用于短语获取\upcite{ja2006obtaining,DBLP:conf/acl/ZhangQMG08}。进一步，研究者提出了更加通用的层次模型来描述翻译过程\upcite{chiang2005a,DBLP:conf/coling/ZollmannVOP08,DBLP:conf/acl/WatanabeTI06}，本章介绍的层次短语模型就是其中典型的代表。之后，使用语言学句法的模型也逐渐兴起。最具代表性的是在单语言端使用语言学句法信息的模型\upcite{DBLP:conf/naacl/GalleyHKM04,galley2006scalable,marcu2006spmt,DBLP:conf/naacl/HuangK06,DBLP:conf/emnlp/DeNeefeKWM07,DBLP:conf/wmt/LiuG08,DBLP:conf/acl/LiuLL06}，即：树到串翻译模型和串到树翻译模型。值得注意的是，除了直接用句法信息定义翻译规则，也有研究者将句法信息作为软约束改进层次短语模型\upcite{zollmann2006syntax,DBLP:conf/acl/MartonR08}。这类方法具有很大的灵活性，既保留了层次短语模型比较健壮的特点，同时也兼顾了语言学句法对翻译的指导作用。在同一时期，也有研究者提出同时使用双语两端的语言学句法树对翻译进行建模，比较有代表性的工作是使用同步树插入文法（Synchronous Tree-Insertion Grammars）和同步树替换文法（Synchronous Tree-Substitution Grammars）进行树到树翻译的建模\upcite{Nesson06inductionof,Zhang07atree-to-tree,DBLP:conf/acl/LiuLL09}。不过，树到树翻译假设两种语言间的句法结构能够相互转换，而这个假设并不总是成立。因此树到树翻译系统往往要配合一些技术，如树二叉化，来提升系统的健壮性。
\vspace{0.5em}
\item 统计机器翻译的另一个基础问题是如何表示并获取翻译单元（如短语）。传统方法中，研究者大多使用词对齐或者句法树等结构化信息，通过启发性方法进行短语和翻译规则的获取。不过这类方法最大的问题是上游系统（比如，词对齐、句法分析等）中的错误会影响到下游系统。因此，很多研究者尝试使用更多样的对齐或者句法分析来指导翻译单元的获取。比如，可以绕过词对齐，直接进行短语对齐\upcite{denero2010phrase}；也可以使用多个句法树或者句法森林来覆盖更多的句法现象，进而增加规则抽取的召回率\upcite{mi2008forest,xiao2010empirical}。另一个有趣的方向是用更紧凑的方式表示更多样的翻译假设，比如，直接将翻译结果用有限状态自动机表示，进行更大搜索空间上的解码\upcite{de2010hierarchical,Casacuberta2004Machine}。
\item 在基于句法的模型中，常常会使用句法分析器完成句法分析树的生成。由于句法分析器会产生错误，因此这些错误会对机器翻译系统产生影响。对于这个问题，一种解决办法是同时考虑更多的句法树，这样增加正确句法分析结果被使用到的概率。其中，比较典型的方式基于句法森林的方法\upcite{DBLP:conf/acl/MiHL08,DBLP:conf/emnlp/MiH08}，比如，在规则抽取或者解码阶段使用句法森林，而不是仅仅使用一棵单独的句法树。另一种思路是，对句法结构进行松弛操作，即在翻译的过程中并不严格遵循句法结构\upcite{DBLP:conf/acl/ZhuX11,DBLP:conf/emnlp/ZhangZZ11}。实际上，前面提到的基于句法软约束的模型也是这类方法的一种体现\upcite{DBLP:conf/wmt/ZollmannV06,DBLP:conf/acl/MartonR08}。实际上，机器翻译领域的长期存在一个问题：使用什么样的句法结构是最适合机器翻译？因此，有研究者尝试对比不同的句法分析结果对机器翻译系统的影响\upcite{DBLP:conf/wmt/PopelMGZ11,DBLP:conf/coling/XiaoZZZ10}。也有研究者面向机器翻译任务自动归纳句法结构\upcite{DBLP:journals/tacl/ZhaiZZZ13}，而不是直接使用从单语小规模树库学习到的句法分析器，这样可以提高系统的健壮性。
\vspace{0.5em}
\item 系统融合是具有统计机器翻译时代特色的研究方向。某种意义上说，系统融合的兴起源于本世纪初各种机器翻译比赛。因为当时提升翻译性能的主要方法之一就是将多个翻译引擎进行融合。系统融合的出发点是：多样的翻译候选有助于生成更好的译文。系统融合有很多思路，比较简单的方法是假设选择，即从多个翻译系统的输出中直接选择一个译文\upcite{bangalore2001computing,rosti2007combining,xiao2013bagging}；另一种方法是用多个系统的输出构建解码格或者混淆网络，这样可以生成新的翻译结果\upcite{Yang2009Lattice,He2008Indirect,Li2009Incremental}；此外，还可以在解码过程中动态融合不同模型\upcite{Yang2009Joint,Mu2009Collaborative}。另一方面，也有研究者探讨如何在一个翻译系统中让不同的模型进行互补，而不是简单的融合。比如，可以控制句法在机器翻译中使用的程度，让句法模型和层次短语模型处理各自擅长的问题\upcite{Tong2016Syntactic}。
\item 本章所讨论的模型大多基于短语结构树。另一个重要的方向是使用依存树进行翻译建模\upcite{DBLP:journals/mt/QuirkM06,DBLP:conf/wmt/XiongLL07,DBLP:conf/coling/Lin04}。依存树比短语结构树有更简单的结构，而且依存关系本身也是对“语义”的表征，因此也可以扑捉到短语结构树所无法涵盖的信息。同其它基于句法的模型类似，基于依存树的模型大多也需要进行规则抽取、解码等步骤，因此这方面的研究工作大多涉及翻译规则的抽取、基于依存树的解码等\upcite{DBLP:conf/acl/DingP05,DBLP:conf/coling/ChenXMJL14,DBLP:conf/coling/SuLMZLL10,DBLP:conf/coling/XieXL14,DBLP:conf/emnlp/LiWL15}。此外，基于依存树的模型也可以与句法森林结构相结合，对系统性能进行进一步提升\upcite{DBLP:conf/acl/MiL10,DBLP:conf/coling/TuLHLL10}。
\vspace{0.5em}
\item 语言模型是统计机器翻译系统所使用的重要特征。但是，即使引入$n$-gram语言模型，机器翻译系统仍然会产生语法上不正确的译文，甚至会生成结构完全错误的译文。对于这个问题，研究者尝试使用基于句法的语言模型。早期的探索有Charniak等人\upcite{charniak2001immediate}和Och等人\upcite{och2004smorgasbord}的工作，不过当时的结果并没有显示出基于句法的语言模型可以显著提升机器翻译的品质。后来，BBN的研究团队提出了基于依存树的语言模型\upcite{shen2008a}，这个模型可以显著提升层次短语模型的性能。正是凭借着这项技术，BBN的系统也在多个机器翻译评测比赛中名列前茅，引起了广泛关注。除此之外，也有研究工作探索基于树替换文法等结构的语言模型\upcite{xiao2011language}。实际上，树到树、串到树模型也可以被看作是一种对目标语言句法合理性的度量，只不过目标语言的句法信息被隐含在翻译规则中。这时，可以在翻译规则上设计相应的特征，以达到引入目标语句法语言模型的目的。
\item 不同模型往往有不同的优点，为了融合这些优点，系统融合是很受关注的研究方向。某种意义上说，系统融合的兴起源于本世纪初各种机器翻译比赛。因为当时提升翻译性能的主要方法之一就是将多个翻译引擎进行融合。系统融合的出发点是：多样的翻译候选有助于生成更好的译文。系统融合的思路很多，比较简单的方法是假设选择（Hypothesis Selection），即从多个翻译系统的输出中直接选择一个译文\upcite{bangalore2001computing,rosti2007combining,xiao2013bagging}；另一种方法是用多个系统的输出构建解码格（Decoding Lattice）或者混淆网络（Confusion Networks），这样可以生成新的翻译结果\upcite{Yang2009Lattice,He2008Indirect,Li2009Incremental}；此外，还可以在解码过程中动态融合不同模型 \upcite{Yang2009Joint,Mu2009Collaborative}。另一方面，也有研究者探讨如何在一个翻译系统中让不同的模型进行互补，而不是简单的融合。比如，可以控制句法在机器翻译中使用的程度，让句法模型和层次短语模型处理各自擅长的问题 \upcite{Tong2016Syntactic}。
\vspace{0.5em}
\item 语言模型是统计机器翻译系统所使用的重要特征。但是，即使引入$n$-gram语言模型，机器翻译系统仍然会产生语法上不正确的译文，甚至会生成结构完全错误的译文。对于这个问题，研究者尝试使用基于句法的语言模型。早期的探索有 Charniak 等人\upcite{charniak2001immediate}和 Och 等人\upcite{och2004smorgasbord}的工作，不过当时的结果并没有显示出基于句法的语言模型可以显著提升机器翻译的品质。后来，BBN 的研究团队提出了基于依存树的语言模型\upcite{shen2008a}，这个模型可以显著提升层次短语模型的性能。除此之外，也有研究工作探索基于树替换文法等结构的语言模型\upcite{xiao2011language}。实际上，树到树、串到树模型也可以被看作是一种对目标语言句法合理性的度量，只不过目标语言的句法信息被隐含在翻译规则中。这时，可以在翻译规则上设计相应的特征，以达到引入目标语句法语言模型的目的。
\vspace{0.5em}
\end{itemize}
......
bibliography.bib
......@@ -3007,12 +3007,14 @@
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2010}
}
@inproceedings{ilprints729,
author = {Dan Klein and
Christopher D. Manning},
title = {Parsing and Hypergraphs},
publisher = {Tsinghua University Press},
year = {2001}
@article{ilprints729,
title={Parsing and Hypergraphs},
author={Klein, Dan and Manning, Christopher D.},
journal={New Developments in Parsing Technology},
volume={65},
number={3},
pages={123--134},
year={2001},
}
@article{goodman1999semiring,
author = {Joshua Goodman},
......@@ -3047,6 +3049,460 @@
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2006}
}
@inproceedings{DBLP:conf/acl/AlshawiBX97,
author = {Hiyan Alshawi and
Adam L. Buchsbaum and
Fei Xia},
title = {A Comparison of Head Transducers and Transfer for a Limited Domain
Translation Application},
pages = {360--365},
publisher = {Morgan Kaufmann Publishers},
year = {1997}
}
@inproceedings{DBLP:conf/acl-vlc/Wu95,
author = {Dekai Wu},
title = {Trainable Coarse Bilingual Grammars for Parallel Text Bracketing},
publisher = {Third Workshop on Very Large Corpor},
year = {1995}
}
@inproceedings{DBLP:conf/acl/WuW98,
author = {Dekai Wu and
Hongsing Wong},
title = {Machine Translation with a Stochastic Grammatical Channel},
pages = {1408--1415},
publisher = {Morgan Kaufmann Publishers},
year = {1998}
}
@inproceedings{ja2006obtaining,
author = {J.A.Sánchez and J.M.Benedí.},
title = {Obtaining Word Phrases with Stochastic Inversion Transduction Grammars for Phrase-based Statistical Machine Translation},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2006}
}
@inproceedings{DBLP:conf/acl/ZhangQMG08,
author = {Hao Zhang and
Chris Quirk and
Robert C. Moore and
Daniel Gildea},
title = {Bayesian Learning of Non-Compositional Phrases with Synchronous Parsing},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2008}
}
@inproceedings{DBLP:conf/coling/ZollmannVOP08,
author = {Andreas Zollmann and
Ashish Venugopal and
Franz Josef Och and
Jay M. Ponte},
title = {A Systematic Comparison of Phrase-Based, Hierarchical and Syntax-Augmented
Statistical {MT}},
publisher = {International Conference on Computational Linguistics},
pages = {1145--1152},
year = {2008}
}
@inproceedings{DBLP:conf/acl/WatanabeTI06,
author = {Taro Watanabe and
Hajime Tsukada and
Hideki Isozaki},
title = {Left-to-Right Target Generation for Hierarchical Phrase-Based Translation},
publisher = {Annual Meeting of the Association for Computational Linguisticss},
year = {2006}
}
@inproceedings{DBLP:conf/naacl/GalleyHKM04,
author = {Michel Galley and
Mark Hopkins and
Kevin Knight and
Daniel Marcu},
title = {What's in a translation rule?},
pages = {273--280},
publisher = {The Association for Computational Linguistics},
year = {2004}
}
@inproceedings{DBLP:conf/naacl/HuangK06,
author = {Bryant Huang and
Kevin Knight},
title = {Relabeling Syntax Trees to Improve Syntax-Based Machine Translation
Quality},
publisher = {The Association for Computational Linguistics},
year = {2006}
}
@inproceedings{DBLP:conf/emnlp/DeNeefeKWM07,
author = {Steve DeNeefe and
Kevin Knight and
Wei Wang and
Daniel Marcu},
title = {What Can Syntax-Based {MT} Learn from Phrase-Based MT?},
pages = {755--763},
publisher = {The Association for Computational Linguistics},
year = {2007}
}
@inproceedings{DBLP:conf/wmt/LiuG08,
author = {Ding Liu and
Daniel Gildea},
title = {Improved Tree-to-String Transducer for Machine Translation},
pages = {62--69},
publisher = {The Association for Computational Linguistics},
year = {2008}
}
@inproceedings{DBLP:conf/acl/LiuLL06,
author = {Yang Liu and
Qun Liu and
Shouxun Lin},
title = {Tree-to-String Alignment Template for Statistical Machine Translation},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2006}
}
@inproceedings{zollmann2006syntax,
author = {Andreas Zollmann and
Ashish Venugopal},
title = {Syntax Augmented Machine Translation via Chart Parsing},
pages = {138--141},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2006}
}
@inproceedings{DBLP:conf/acl/MartonR08,
author = {Yuval Marton and
Philip Resnik},
title = {Soft Syntactic Constraints for Hierarchical Phrased-Based Translation},
pages = {1003--1011},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2008}
}
@INPROCEEDINGS{Nesson06inductionof,
author = {Rebecca Nesson and Stuart M. Shieber and Alexander Rush},
title = {Induction of probabilistic synchronous tree-insertion grammars for machine translation},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2006}
}
@MISC{Zhang07atree-to-tree,
author = {Min Zhang and Hongfei Jiang and Ai Ti Aw and Jun Sun and Sheng Li and Chew Lim Tan},
title = {A Tree-to-Tree Alignment-based Model for Statistical Machine Translation},
year = {2007},
publisher = {MT-Summit}
}
@inproceedings{DBLP:conf/acl/LiuLL09,
author = {Yang Liu and
Yajuan L{\"{u}} and
Qun Liu},
title = {Improving Tree-to-Tree Translation with Packed Forests},
pages = {558--566},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2009}
}
@inproceedings{DBLP:conf/emnlp/WangKM07,
author = {Wei Wang and
Kevin Knight and
Daniel Marcu},
title = {Binarizing Syntax Trees to Improve Syntax-Based Machine Translation
Accuracy},
pages = {746--754},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2007}
}
@inproceedings{DBLP:conf/acl/MiHL08,
author = {Haitao Mi and
Liang Huang and
Qun Liu},
title = {Forest-Based Translation},
pages = {192--199},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2008}
}
@inproceedings{DBLP:conf/emnlp/MiH08,
author = {Haitao Mi and
Liang Huang},
title = {Forest-based Translation Rule Extraction},
pages = {206--214},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2008}
}
@inproceedings{DBLP:conf/acl/ZhuX11,
author = {Jingbo Zhu and
Tong Xiao},
title = {Improving Decoding Generalization for Tree-to-String Translation},
pages = {418--423},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2011}
}
@inproceedings{DBLP:conf/emnlp/ZhangZZ11,
author = {Jiajun Zhang and
Feifei Zhai and
Chengqing Zong},
title = {Augmenting String-to-Tree Translation Models with Fuzzy Use of Source-side
Syntax},
pages = {204--215},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2011}
}
@inproceedings{DBLP:conf/wmt/ZollmannV06,
author = {Andreas Zollmann and
Ashish Venugopal},
title = {Syntax Augmented Machine Translation via Chart Parsing},
pages = {138--141},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2006}
}
@inproceedings{DBLP:conf/acl/MartonR08,
author = {Yuval Marton and
Philip Resnik},
title = {Soft Syntactic Constraints for Hierarchical Phrased-Based Translation},
pages = {1003--1011},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2008}
}
@inproceedings{DBLP:conf/wmt/PopelMGZ11,
author = {Martin Popel and
David Marecek and
Nathan Green and
Zdenek Zabokrtsk{\'{y}}},
title = {Influence of Parser Choice on Dependency-Based {MT}},
pages = {433--439},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2011}
}
@inproceedings{DBLP:conf/coling/XiaoZZZ10,
author = {Tong Xiao and
Jingbo Zhu and
Hao Zhang and
Muhua Zhu},
title = {An Empirical Study of Translation Rule Extraction with Multiple Parsers},
pages = {1345--1353},
publisher = {Chinese Information Processing Society of China},
year = {2010}
}
@article{Zhai2012Treebased,
title={Treebased translation without using parse trees},
author={Zhai, Feifei and Zhang, Jiajun and Zhou, Yu and Zong, Chengqing},
publisher = {International Conference on Computational Linguistics},
year={2012},
}
@article{DBLP:journals/tacl/ZhaiZZZ13,
author = {Feifei Zhai and
Jiajun Zhang and
Yu Zhou and
Chengqing Zong},
title = {Unsupervised Tree Induction for Tree-based Translation},
journal = {Trans. Assoc. Comput. Linguistics},
volume = {1},
pages = {243--254},
year = {2013}
}
@article{DBLP:journals/mt/QuirkM06,
author = {Christopher Quirk and
Arul Menezes},
title = {Dependency treelet translation: the convergence of statistical and
example-based machine-translation?},
journal = {Machine Translation},
volume = {20},
number = {1},
pages = {43--65},
year = {2006}
}
@inproceedings{DBLP:conf/wmt/XiongLL07,
author = {Deyi Xiong and
Qun Liu and
Shouxun Lin},
title = {A Dependency Treelet String Correspondence Model for Statistical Machine
Translation},
pages = {40--47},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2007}
}
@inproceedings{DBLP:conf/coling/Lin04,
author = {Dekang Lin},
title = {A Path-based Transfer Model for Machine Translation},
publisher = {International Conference on Computational Linguistics},
year = {2004}
}
@inproceedings{DBLP:conf/acl/DingP05,
author = {Yuan Ding and
Martha Palmer},
title = {Machine Translation Using Probabilistic Synchronous Dependency Insertion
Grammars},
pages = {541--548},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2005}
}
@inproceedings{DBLP:conf/coling/ChenXMJL14,
author = {Hongshen Chen and
Jun Xie and
Fandong Meng and
Wenbin Jiang and
Qun Liu},
title = {A Dependency Edge-based Transfer Model for Statistical Machine Translation},
pages = {1103--1113},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2014}
}
@inproceedings{DBLP:conf/coling/SuLMZLL10,
author = {Jinsong Su and
Yang Liu and
Haitao Mi and
Hongmei Zhao and
Yajuan Lv and
Qun Liu},
title = {Dependency-Based Bracketing Transduction Grammar for Statistical Machine
Translation},
pages = {1185--1193},
publisher = {Chinese Information Processing Society of China},
year = {2010}
}
@inproceedings{DBLP:conf/coling/XieXL14,
author = {Jun Xie and
Jinan Xu and
Qun Liu},
title = {Augment Dependency-to-String Translation with Fixed and Floating Structures},
pages = {2217--2226},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2014}
}
@inproceedings{DBLP:conf/emnlp/LiWL15,
author = {Liangyou Li and
Andy Way and
Qun Liu},
title = {Dependency Graph-to-String Translation},
pages = {33--43},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2015}
}
@inproceedings{DBLP:conf/acl/MiL10,
author = {Haitao Mi and
Qun Liu},
title = {Constituency to Dependency Translation with Forests},
pages = {1433--1442},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2010}
}
@inproceedings{DBLP:conf/coling/TuLHLL10,
author = {Zhaopeng Tu and
Yang Liu and
Young{-}Sook Hwang and
Qun Liu and
Shouxun Lin},
title = {Dependency Forest for Statistical Machine Translation},
pages = {1092--1100},
publisher = {International Conference on Computational Linguistics},
year = {2010}
}
@inproceedings{bangalore2001computing,
title ={Computing consensus translation from multiple machine translation systems},
author ={Bangalore, B and Bordel, German and Riccardi, Giuseppe},
pages ={351--354},
year ={2001},
organization ={The Institute of Electrical and Electronics Engineers}
}
@inproceedings{rosti2007combining,
author = {Antti{-}Veikko I. Rosti and
Necip Fazil Ayan and
Bing Xiang and
Spyridon Matsoukas and
Richard M. Schwartz and
Bonnie J. Dorr},
title = {Combining Outputs from Multiple Machine Translation Systems},
pages = {228--235},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2007}
}
@article{xiao2013bagging,
title ={Bagging and boosting statistical machine translation systems},
author ={Xiao, Tong and Zhu, Jingbo and Liu, Tongran},
publisher ={Artificial Intelligence},
volume ={195},
pages ={496--527},
year ={2013}
}
@inproceedings{Yang2009Lattice,
author = {Yang Feng and
Yang Liu and
Haitao Mi and
Qun Liu and
Yajuan L{\"{u}}},
title = {Lattice-based System Combination for Statistical Machine Translation},
pages = {1105--1113},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2009}
}
@inproceedings{He2008Indirect,
author = {Xiaodong He and
Mei Yang and
Jianfeng Gao and
Patrick Nguyen and
Robert Moore},
title = {Indirect-HMM-based Hypothesis Alignment for Combining Outputs from
Machine Translation Systems},
pages = {98--107},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2008}
}
@inproceedings{Li2009Incremental,
author = {Chi{-}Ho Li and
Xiaodong He and
Yupeng Liu and
Ning Xi},
title = {Incremental {HMM} Alignment for {MT} System Combination},
pages = {949--957},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2009}
}
@inproceedings{Yang2009Joint,
author = {Yang Liu and
Haitao Mi and
Yang Feng and
Qun Liu},
title = {Joint Decoding with Multiple Translation Models},
pages = {576--584},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2009}
}
@inproceedings{Mu2009Collaborative,
author = {Mu Li and
Nan Duan and
Dongdong Zhang and
Chi{-}Ho Li and
Ming Zhou},
title = {Collaborative Decoding: Partial Hypothesis Re-ranking Using Translation
Consensus between Decoders},
pages = {585--592},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2009}
}
@inproceedings{Tong2016Syntactic,
author = {Tong Xiao and
Jingbo Zhu and
Chunliang Zhang and
Tongran Liu},
title = {Syntactic Skeleton-Based Translation},
pages = {2856--2862},
publisher = {AAAI Conference on Artificial Intelligence},
year = {2016},
}
@inproceedings{charniak2001immediate,
author = {Eugene Charniak},
title = {Immediate-Head Parsing for Language Models},
pages = {116--123},
publisher = {Morgan Kaufmann Publishers},
year = {2001}
}
@inproceedings{shen2008a,
author = {Libin Shen and
Jinxi Xu and
Ralph M. Weischedel},
title = {A New String-to-Dependency Machine Translation Algorithm with a Target
Dependency Language Model},
pages = {577--585},
publisher = {Annual Meeting of the Association for Computational Linguistics},
year = {2008}
}
@article{xiao2011language,
title ={Language Modeling for Syntax-Based Machine Translation Using Tree Substitution Grammars: A Case Study on Chinese-English Translation},
author ={Xiao, Tong and Zhu, Jingbo and Zhu, Muhua},
volume ={10},
number ={4},
pages ={1--29},
year ={2011},
publisher ={ACM Transactions on Asian Language Information Processing (TALIP)}
}
%%%%% chapter 8------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
......
Markdown 格式
0% 或
您添加了 0 人 到此讨论。请谨慎行事。
请先完成此评论的编辑！
请 注册 或者 后发表评论