[ Skip to the content ]

Institute of Formal and Applied Linguistics

at Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic


[ Back to the navigation ]

Publication


Year 2011
Type article
Status published
Language English
Author(s) Mareček, David
Title Combining Diverse Word-Alignment Symmetrizations Improves Dependency Tree Projection
Czech title Kombinace různých symetrizací slovního zarovnání zlepšuje projekci závislostních stromů
Journal Lecture Notes in Computer Science
Publisher's city and country Berlin / Heidelberg
Volume 6608
Pages range 144-154
Month February
URL http://www.springerlink.com/content/978-3-642-19399-6/#section=855851&page=1&locus=0
Supported by 2010-2013 FP7-ICT-2009-4-247762 (Faust) 2009-2012 GD201/09/H057 (Res Informatica)
Czech abstract Pro mnoho jazyků nemůžeme natrénovat parser, protože nemáme k dispozici žádná ručně anotovaná data. Tento problém můžeme vyřešit použitím paralelního korpusu s angličtinou naparsováním amglické strany a projekcí závislostí do druhého jazyka.
English abstract For many languages, we are not able to train any supervised parser, because there are no manually annotated data available. This problem can be solved by using a parallel corpus with English, parsing the English side, projecting the dependencies through word-alignment connections, and training a parser on the projected trees. In this paper, we introduce a simple algorithm using a combination of various word-alignment symmetrizations. We prove that our method outperforms previous work, even though it uses McDonald's maximum-spanning-tree parser as it is, without any "unsupervised" modifications.
Specialization linguistics ("jazykověda")
Confidentiality default – not confidential
Open access no
WOS Code 000302390500012
DOI 10.1007/978-3-642-19400-9
ISSN* 0302-9743
Publisher* Springer Verlag Heidelberg
Creator: Common Account
Created: 10/31/11 9:58 AM
Modifier: Common Account
Modified: 9/8/15 11:05 AM
***

Content, Design & Functionality: ÚFAL, 2006–2016. Page generated: Mon Oct 15 18:18:44 CEST 2018

[ Back to the navigation ] [ Back to the content ]

100% OpenAIRE compliant