Transition-Based Deep Input Linearization.

Ratish Puduppully; Yue Zhang; Manish Shrivastava

doi:10.18653/V1/E17-1061

Transition-Based Deep Input Linearization.

Ratish Puduppully, Yue Zhang, Manish Shrivastava

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

Abstract

Traditional methods for deep NLG adopt pipeline approaches comprising stages such as constructing syntactic input, predicting function words, linearizing the syntactic input and generating the surface forms. Though easier to visualize, pipeline approaches suffer from error propagation. In addition, information available across modules cannot be leveraged by all modules. We construct a transition-based model to jointly perform linearization, function word prediction and morphological generation, which considerably improves upon the accuracy compared to a pipelined baseline system. On a standard deep input linearization shared task, our system achieves the best results reported so far.

Original language	English
Title of host publication	Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers
Publisher	Association for Computational Linguistics
Publication date	2017
Pages	643-654
DOIs	https://doi.org/10.18653/V1/E17-1061
Publication status	Published - 2017
Externally published	Yes

Access to Document

10.18653/V1/E17-1061

https://dblp.org/rec/conf/eacl/ZhangSP17

Cite this

@inproceedings{ea3d895a516d4734828ea084185d1231,

title = "Transition-Based Deep Input Linearization.",

abstract = "Traditional methods for deep NLG adopt pipeline approaches comprising stages such as constructing syntactic input, predicting function words, linearizing the syntactic input and generating the surface forms. Though easier to visualize, pipeline approaches suffer from error propagation. In addition, information available across modules cannot be leveraged by all modules. We construct a transition-based model to jointly perform linearization, function word prediction and morphological generation, which considerably improves upon the accuracy compared to a pipelined baseline system. On a standard deep input linearization shared task, our system achieves the best results reported so far.",

author = "Ratish Puduppully and Yue Zhang and Manish Shrivastava",

note = "DBLP's bibliographic metadata records provided through http://dblp.org/search/publ/api are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.",

year = "2017",

doi = "10.18653/V1/E17-1061",

language = "English",

pages = "643--654",

booktitle = "Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers",

publisher = "Association for Computational Linguistics",

address = "United States",

}

Transition-Based Deep Input Linearization. / Puduppully, Ratish; Zhang, Yue; Shrivastava, Manish.
Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers. Association for Computational Linguistics, 2017. p. 643-654.

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

TY - GEN

T1 - Transition-Based Deep Input Linearization.

AU - Puduppully, Ratish

AU - Zhang, Yue

AU - Shrivastava, Manish

N1 - DBLP's bibliographic metadata records provided through http://dblp.org/search/publ/api are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.

PY - 2017

Y1 - 2017

N2 - Traditional methods for deep NLG adopt pipeline approaches comprising stages such as constructing syntactic input, predicting function words, linearizing the syntactic input and generating the surface forms. Though easier to visualize, pipeline approaches suffer from error propagation. In addition, information available across modules cannot be leveraged by all modules. We construct a transition-based model to jointly perform linearization, function word prediction and morphological generation, which considerably improves upon the accuracy compared to a pipelined baseline system. On a standard deep input linearization shared task, our system achieves the best results reported so far.

AB - Traditional methods for deep NLG adopt pipeline approaches comprising stages such as constructing syntactic input, predicting function words, linearizing the syntactic input and generating the surface forms. Though easier to visualize, pipeline approaches suffer from error propagation. In addition, information available across modules cannot be leveraged by all modules. We construct a transition-based model to jointly perform linearization, function word prediction and morphological generation, which considerably improves upon the accuracy compared to a pipelined baseline system. On a standard deep input linearization shared task, our system achieves the best results reported so far.

U2 - 10.18653/V1/E17-1061

DO - 10.18653/V1/E17-1061

M3 - Article in proceedings

SP - 643

EP - 654

BT - Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers

PB - Association for Computational Linguistics

ER -

Transition-Based Deep Input Linearization.

Abstract

Access to Document

Fingerprint

Cite this