A system for multilingual dependency parsing based on bidirectional LSTM feature representations

Kyung Tae Lim; Thierry Poibeau

doi:10.18653/v1/k17-3006

A system for multilingual dependency parsing based on bidirectional LSTM feature representations

Kyung Tae Lim, Thierry Poibeau

Dept. of Applied Artificial Intelligence

École normale supérieure

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

9 Scopus citations

Abstract

In this paper, we present our multilingual dependency parser developed for the CoNLL 2017 UD Shared Task dealing with “Multilingual Parsing from Raw Text to Universal Dependencies”¹. Our parser extends the monolingual BIST-parser as a multi-source multilingual trainable parser. Thanks to multilingual word embeddings and one hot encodings for languages, our system can use both monolingual and multi-source training. We trained 69 monolingual language models and 13 multilingual models for the shared task. Our multilingual approach making use of different resources yield better results than the monolingual approach for 11 languages. Our system ranked 5^th and achieved 70.93 overall LAS score over the 81 test corpora (macro-averaged LAS F1 score).

Original language	English
Title of host publication	CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task
Subtitle of host publication	Multilingual Parsing from Raw Text to Universal Dependencies
Publisher	Association for Computational Linguistics (ACL)
Pages	63-70
Number of pages	8
ISBN (Electronic)	9781945626708
DOIs	https://doi.org/10.18653/v1/k17-3006
State	Published - 2017
Event	2017 SIGNLL Conference on Computational Natural Language Learning- CoNLL Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, CoNLL 2017 - Vancouver, Canada Duration: 3 Aug 2017 → 4 Aug 2017

Publication series

Name	CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies

Conference

Conference	2017 SIGNLL Conference on Computational Natural Language Learning- CoNLL Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, CoNLL 2017
Country/Territory	Canada
City	Vancouver
Period	3/08/17 → 4/08/17

Access to Document

10.18653/v1/k17-3006

Cite this

Lim, K. T., & Poibeau, T. (2017). A system for multilingual dependency parsing based on bidirectional LSTM feature representations. In CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies (pp. 63-70). (CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies). Association for Computational Linguistics (ACL). https://doi.org/10.18653/v1/k17-3006

Lim, Kyung Tae ; Poibeau, Thierry. / A system for multilingual dependency parsing based on bidirectional LSTM feature representations. CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies. Association for Computational Linguistics (ACL), 2017. pp. 63-70 (CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies).

@inproceedings{00ffeb0db4ef4562bcf16f26b954e89c,

title = "A system for multilingual dependency parsing based on bidirectional LSTM feature representations",

abstract = "In this paper, we present our multilingual dependency parser developed for the CoNLL 2017 UD Shared Task dealing with “Multilingual Parsing from Raw Text to Universal Dependencies”1. Our parser extends the monolingual BIST-parser as a multi-source multilingual trainable parser. Thanks to multilingual word embeddings and one hot encodings for languages, our system can use both monolingual and multi-source training. We trained 69 monolingual language models and 13 multilingual models for the shared task. Our multilingual approach making use of different resources yield better results than the monolingual approach for 11 languages. Our system ranked 5th and achieved 70.93 overall LAS score over the 81 test corpora (macro-averaged LAS F1 score).",

author = "Lim, \{Kyung Tae\} and Thierry Poibeau",

note = "Publisher Copyright: {\textcopyright} 2017 Association for Computational Linguistics.; 2017 SIGNLL Conference on Computational Natural Language Learning- CoNLL Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, CoNLL 2017 ; Conference date: 03-08-2017 Through 04-08-2017",

year = "2017",

doi = "10.18653/v1/k17-3006",

language = "English",

series = "CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies",

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

pages = "63--70",

booktitle = "CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task",

}

Lim, KT & Poibeau, T 2017, A system for multilingual dependency parsing based on bidirectional LSTM feature representations. in CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies. CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, Association for Computational Linguistics (ACL), pp. 63-70, 2017 SIGNLL Conference on Computational Natural Language Learning- CoNLL Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, CoNLL 2017, Vancouver, Canada, 3/08/17. https://doi.org/10.18653/v1/k17-3006

A system for multilingual dependency parsing based on bidirectional LSTM feature representations. / Lim, Kyung Tae; Poibeau, Thierry.
CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies. Association for Computational Linguistics (ACL), 2017. p. 63-70 (CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies).

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

TY - GEN

T1 - A system for multilingual dependency parsing based on bidirectional LSTM feature representations

AU - Lim, Kyung Tae

AU - Poibeau, Thierry

PY - 2017

Y1 - 2017

N2 - In this paper, we present our multilingual dependency parser developed for the CoNLL 2017 UD Shared Task dealing with “Multilingual Parsing from Raw Text to Universal Dependencies”1. Our parser extends the monolingual BIST-parser as a multi-source multilingual trainable parser. Thanks to multilingual word embeddings and one hot encodings for languages, our system can use both monolingual and multi-source training. We trained 69 monolingual language models and 13 multilingual models for the shared task. Our multilingual approach making use of different resources yield better results than the monolingual approach for 11 languages. Our system ranked 5th and achieved 70.93 overall LAS score over the 81 test corpora (macro-averaged LAS F1 score).

AB - In this paper, we present our multilingual dependency parser developed for the CoNLL 2017 UD Shared Task dealing with “Multilingual Parsing from Raw Text to Universal Dependencies”1. Our parser extends the monolingual BIST-parser as a multi-source multilingual trainable parser. Thanks to multilingual word embeddings and one hot encodings for languages, our system can use both monolingual and multi-source training. We trained 69 monolingual language models and 13 multilingual models for the shared task. Our multilingual approach making use of different resources yield better results than the monolingual approach for 11 languages. Our system ranked 5th and achieved 70.93 overall LAS score over the 81 test corpora (macro-averaged LAS F1 score).

UR - https://www.scopus.com/pages/publications/85059887667

U2 - 10.18653/v1/k17-3006

DO - 10.18653/v1/k17-3006

M3 - Conference contribution

AN - SCOPUS:85059887667

T3 - CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies

SP - 63

EP - 70

BT - CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task

PB - Association for Computational Linguistics (ACL)

T2 - 2017 SIGNLL Conference on Computational Natural Language Learning- CoNLL Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, CoNLL 2017

Y2 - 3 August 2017 through 4 August 2017

ER -

Lim KT, Poibeau T. A system for multilingual dependency parsing based on bidirectional LSTM feature representations. In CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies. Association for Computational Linguistics (ACL). 2017. p. 63-70. (CoNLL 2017 - SIGNLL Conference on Computational Natural Language Learning, Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies). doi: 10.18653/v1/k17-3006

A system for multilingual dependency parsing based on bidirectional LSTM feature representations

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this