Machine comprehension improves domain-specific japanese predicate-argument structure analysis

Norio Takahashi; Tomohide Shibata; Daisuke Kawahara; Sadao Kurohashi

Machine comprehension improves domain-specific japanese predicate-argument structure analysis

Norio Takahashi, Tomohide Shibata, Daisuke Kawahara, Sadao Kurohashi

研究成果: Conference contribution

抄録

To improve the accuracy of predicate- argument structure (PAS) analysis, large-scale training data and knowledge for PAS analysis are indispensable. We focus on a specific domain, specifically Japanese blogs on driv- ing, and construct two wide-coverage datasets as a form of QA using crowdsourcing: A PAS-QA dataset and a reading comprehension QA (RC-QA) dataset. We train a machine comprehension (MC) model based on these datasets to perform PAS analysis. Our experi- ments show that a stepwise training method is the most effective, which pre-trains an MC model based on the RC-QA dataset to acquire domain knowledge and then fine-tunes based on the PAS-QA dataset.

本文言語	English
ホスト出版物のタイトル	MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering
出版社	Association for Computational Linguistics (ACL)
ページ	98-104
ページ数	7
ISBN（電子版）	9781950737819
出版ステータス	Published - 2019
外部発表	はい
イベント	2nd Workshop on Machine Reading for Question Answering, MRQA@EMNLP 2019 - Hong Kong, China 継続期間: 2019 11月 4 → …

出版物シリーズ

名前	MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering

Conference

Conference	2nd Workshop on Machine Reading for Question Answering, MRQA@EMNLP 2019
国/地域	China
City	Hong Kong
Period	19/11/4 → …

ASJC Scopus subject areas

コンピュータグラフィックスおよびコンピュータ支援設計
人工知能

他のファイルとリンク

引用スタイル

Takahashi, N., Shibata, T., Kawahara, D., & Kurohashi, S. (2019). Machine comprehension improves domain-specific japanese predicate-argument structure analysis. In MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering (pp. 98-104). (MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering). Association for Computational Linguistics (ACL).

Machine comprehension improves domain-specific japanese predicate-argument structure analysis. / Takahashi, Norio; Shibata, Tomohide; Kawahara, Daisuke その他.

MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering. Association for Computational Linguistics (ACL), 2019. p. 98-104 (MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering).

研究成果: Conference contribution

Takahashi, N, Shibata, T, Kawahara, D & Kurohashi, S 2019, Machine comprehension improves domain-specific japanese predicate-argument structure analysis. in MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering. MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering, Association for Computational Linguistics (ACL), pp. 98-104, 2nd Workshop on Machine Reading for Question Answering, MRQA@EMNLP 2019, Hong Kong, China, 19/11/4.

Takahashi N, Shibata T, Kawahara D, Kurohashi S. Machine comprehension improves domain-specific japanese predicate-argument structure analysis. In MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering. Association for Computational Linguistics (ACL). 2019. p. 98-104. (MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering).

Takahashi, Norio ; Shibata, Tomohide ; Kawahara, Daisuke その他. / Machine comprehension improves domain-specific japanese predicate-argument structure analysis. MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering. Association for Computational Linguistics (ACL), 2019. pp. 98-104 (MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering).

@inproceedings{67ac7b50f1a14bf8b5b8be321b25469e,

title = "Machine comprehension improves domain-specific japanese predicate-argument structure analysis",

abstract = "To improve the accuracy of predicate- argument structure (PAS) analysis, large-scale training data and knowledge for PAS analysis are indispensable. We focus on a specific domain, specifically Japanese blogs on driv- ing, and construct two wide-coverage datasets as a form of QA using crowdsourcing: A PAS-QA dataset and a reading comprehension QA (RC-QA) dataset. We train a machine comprehension (MC) model based on these datasets to perform PAS analysis. Our experi- ments show that a stepwise training method is the most effective, which pre-trains an MC model based on the RC-QA dataset to acquire domain knowledge and then fine-tunes based on the PAS-QA dataset.",

author = "Norio Takahashi and Tomohide Shibata and Daisuke Kawahara and Sadao Kurohashi",

note = "Publisher Copyright: {\textcopyright} 2019 MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering. All rights reserved.; 2nd Workshop on Machine Reading for Question Answering, MRQA@EMNLP 2019 ; Conference date: 04-11-2019",

year = "2019",

language = "English",

series = "MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering",

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

pages = "98--104",

booktitle = "MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering",

}

TY - GEN

T1 - Machine comprehension improves domain-specific japanese predicate-argument structure analysis

AU - Takahashi, Norio

AU - Shibata, Tomohide

AU - Kawahara, Daisuke

AU - Kurohashi, Sadao

PY - 2019

Y1 - 2019

N2 - To improve the accuracy of predicate- argument structure (PAS) analysis, large-scale training data and knowledge for PAS analysis are indispensable. We focus on a specific domain, specifically Japanese blogs on driv- ing, and construct two wide-coverage datasets as a form of QA using crowdsourcing: A PAS-QA dataset and a reading comprehension QA (RC-QA) dataset. We train a machine comprehension (MC) model based on these datasets to perform PAS analysis. Our experi- ments show that a stepwise training method is the most effective, which pre-trains an MC model based on the RC-QA dataset to acquire domain knowledge and then fine-tunes based on the PAS-QA dataset.

AB - To improve the accuracy of predicate- argument structure (PAS) analysis, large-scale training data and knowledge for PAS analysis are indispensable. We focus on a specific domain, specifically Japanese blogs on driv- ing, and construct two wide-coverage datasets as a form of QA using crowdsourcing: A PAS-QA dataset and a reading comprehension QA (RC-QA) dataset. We train a machine comprehension (MC) model based on these datasets to perform PAS analysis. Our experi- ments show that a stepwise training method is the most effective, which pre-trains an MC model based on the RC-QA dataset to acquire domain knowledge and then fine-tunes based on the PAS-QA dataset.

UR - http://www.scopus.com/inward/record.url?scp=85108648722&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85108648722&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85108648722

T3 - MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering

SP - 98

EP - 104

BT - MRQA@EMNLP 2019 - Proceedings of the 2nd Workshop on Machine Reading for Question Answering

PB - Association for Computational Linguistics (ACL)

T2 - 2nd Workshop on Machine Reading for Question Answering, MRQA@EMNLP 2019

Y2 - 4 November 2019

ER -

Machine comprehension improves domain-specific japanese predicate-argument structure analysis

抄録

出版物シリーズ

Conference

ASJC Scopus subject areas

他のファイルとリンク

フィンガープリント

引用スタイル