Dynamic feature selection for dependency parsing

He He; Hal Daumé; Jason Eisner

Dynamic feature selection for dependency parsing

He He, Hal Daumé, Jason Eisner

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Feature computation and exhaustive search have significantly restricted the speed of graph-based dependency parsing. We propose a faster framework of dynamic feature selection, where features are added sequentially as needed, edges are pruned early, and decisions are made online for each sentence. We model this as a sequential decision-making problem and solve it by imitation learning techniques. We test our method on 7 languages. Our dynamic parser can achieve accuracies comparable or even superior to parsers using a full set of features, while computing fewer than 30% of the feature templates.

Original language	English (US)
Title of host publication	EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference
Publisher	Association for Computational Linguistics (ACL)
Pages	1455-1464
Number of pages	10
ISBN (Electronic)	9781937284978
State	Published - 2013
Event	2013 Conference on Empirical Methods in Natural Language Processing, EMNLP 2013 - Seattle, United States Duration: Oct 18 2013 → Oct 21 2013

Publication series

Name	EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference

Other

Other	2013 Conference on Empirical Methods in Natural Language Processing, EMNLP 2013
Country/Territory	United States
City	Seattle
Period	10/18/13 → 10/21/13

ASJC Scopus subject areas

Computational Theory and Mathematics
Information Systems
Computer Vision and Pattern Recognition

Cite this

He, H., Daumé, H., & Eisner, J. (2013). Dynamic feature selection for dependency parsing. In EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference (pp. 1455-1464). (EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference). Association for Computational Linguistics (ACL).

Dynamic feature selection for dependency parsing. / He, He; Daumé, Hal; Eisner, Jason.
EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference. Association for Computational Linguistics (ACL), 2013. p. 1455-1464 (EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

He, H, Daumé, H & Eisner, J 2013, Dynamic feature selection for dependency parsing. in EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference. EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference, Association for Computational Linguistics (ACL), pp. 1455-1464, 2013 Conference on Empirical Methods in Natural Language Processing, EMNLP 2013, Seattle, United States, 10/18/13.

He H, Daumé H, Eisner J. Dynamic feature selection for dependency parsing. In EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference. Association for Computational Linguistics (ACL). 2013. p. 1455-1464. (EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference).

He, He ; Daumé, Hal ; Eisner, Jason. / Dynamic feature selection for dependency parsing. EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference. Association for Computational Linguistics (ACL), 2013. pp. 1455-1464 (EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference).

@inproceedings{158bd8de848b4c6c801b05ddb3b3a348,

title = "Dynamic feature selection for dependency parsing",

abstract = "Feature computation and exhaustive search have significantly restricted the speed of graph-based dependency parsing. We propose a faster framework of dynamic feature selection, where features are added sequentially as needed, edges are pruned early, and decisions are made online for each sentence. We model this as a sequential decision-making problem and solve it by imitation learning techniques. We test our method on 7 languages. Our dynamic parser can achieve accuracies comparable or even superior to parsers using a full set of features, while computing fewer than 30% of the feature templates.",

author = "He He and Hal Daum{\'e} and Jason Eisner",

note = "Publisher Copyright: {\textcopyright} 2013 Association for Computational Linguistics.; 2013 Conference on Empirical Methods in Natural Language Processing, EMNLP 2013 ; Conference date: 18-10-2013 Through 21-10-2013",

year = "2013",

language = "English (US)",

series = "EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference",

publisher = "Association for Computational Linguistics (ACL)",

pages = "1455--1464",

booktitle = "EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference",

}

TY - GEN

T1 - Dynamic feature selection for dependency parsing

AU - He, He

AU - Daumé, Hal

AU - Eisner, Jason

PY - 2013

Y1 - 2013

N2 - Feature computation and exhaustive search have significantly restricted the speed of graph-based dependency parsing. We propose a faster framework of dynamic feature selection, where features are added sequentially as needed, edges are pruned early, and decisions are made online for each sentence. We model this as a sequential decision-making problem and solve it by imitation learning techniques. We test our method on 7 languages. Our dynamic parser can achieve accuracies comparable or even superior to parsers using a full set of features, while computing fewer than 30% of the feature templates.

AB - Feature computation and exhaustive search have significantly restricted the speed of graph-based dependency parsing. We propose a faster framework of dynamic feature selection, where features are added sequentially as needed, edges are pruned early, and decisions are made online for each sentence. We model this as a sequential decision-making problem and solve it by imitation learning techniques. We test our method on 7 languages. Our dynamic parser can achieve accuracies comparable or even superior to parsers using a full set of features, while computing fewer than 30% of the feature templates.

UR - http://www.scopus.com/inward/record.url?scp=84926348127&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84926348127&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84926348127

T3 - EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference

SP - 1455

EP - 1464

BT - EMNLP 2013 - 2013 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference

PB - Association for Computational Linguistics (ACL)

T2 - 2013 Conference on Empirical Methods in Natural Language Processing, EMNLP 2013

Y2 - 18 October 2013 through 21 October 2013

ER -

Dynamic feature selection for dependency parsing

Abstract

Publication series

Other

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this