Studying the Impact of Filling Information Gaps on the Output Quality of Neural Data-to-Text

Craig Alexander Thomson; Zhijie  Zhao; Somayajulu Gowri Sripada

Studying the Impact of Filling Information Gaps on the Output Quality of Neural Data-to-Text

Craig Alexander Thomson, Zhijie Zhao, Somayajulu Gowri Sripada

University of Aberdeen

Research output: Contribution to conference › Unpublished paper › peer-review

3 Citations (Scopus)

6 Downloads (Pure)

Abstract

It is unfair to expect neural data-to-text to produce high quality output when there are gaps between system input data and information contained in the training text. Thomson et al. (2020) identify and narrow information gaps in Rotowire, a popular data-to-text dataset. In this paper, we describe a study which finds that a state-of-the-art neural data-to-text system produces higher quality output, according
to the information extraction (IE) based metrics, when additional input data is carefully selected from this newly available source. It remains to be shown, however, whether IE metrics used in this study correlate well with humans in judging text quality

Original language	English
Pages	35-40
Number of pages	6
Publication status	Published - Dec 2020
Event	Proceedings of the 13th International Conference on Natural Language Generation - Held online Dublin City University, Dublin, Ireland Duration: 15 Dec 2020 → 18 Dec 2020 Conference number: 13 https://www.inlg2020.org/

Conference

Conference	Proceedings of the 13th International Conference on Natural Language Generation
Abbreviated title	INLG 2020
Country/Territory	Ireland
City	Dublin
Period	15/12/20 → 18/12/20
Internet address	https://www.inlg2020.org/

Bibliographical note

Acknowledgments
We would like to thank our reviewers for their insightful feedback and questions.
The work presented here is partially funded by the Engineering and Physical Sciences Research Council (EPSRC), which funds Craig Thomson under a National Productivity Investment Fund Doctoral Studentship (EP/R512412/1).

Access to Document

Thomson_et_al_ACL_StudyingTheImpact_VoR
Materials published in or after 2016 are licensed on a Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/
Final published version, 193 KBLicence: CC BY

https://www.aclweb.org/anthology/2020.inlg-1.6/Licence: CC BY

Cite this

Studying the Impact of Filling Information Gaps on the Output Quality of Neural Data-to-Text. / Thomson, Craig Alexander; Zhao, Zhijie ; Sripada, Somayajulu Gowri.
2020. 35-40 Paper presented at Proceedings of the 13th International Conference on Natural Language Generation, Dublin, Ireland.

Research output: Contribution to conference › Unpublished paper › peer-review

@conference{07fa7bc736a74e8585196a8305535cf3,

title = "Studying the Impact of Filling Information Gaps on the Output Quality of Neural Data-to-Text",

abstract = "It is unfair to expect neural data-to-text to produce high quality output when there are gaps between system input data and information contained in the training text. Thomson et al. (2020) identify and narrow information gaps in Rotowire, a popular data-to-text dataset. In this paper, we describe a study which finds that a state-of-the-art neural data-to-text system produces higher quality output, accordingto the information extraction (IE) based metrics, when additional input data is carefully selected from this newly available source. It remains to be shown, however, whether IE metrics used in this study correlate well with humans in judging text quality",

author = "Thomson, {Craig Alexander} and Zhijie Zhao and Sripada, {Somayajulu Gowri}",

note = "Acknowledgments We would like to thank our reviewers for their insightful feedback and questions. The work presented here is partially funded by the Engineering and Physical Sciences Research Council (EPSRC), which funds Craig Thomson under a National Productivity Investment Fund Doctoral Studentship (EP/R512412/1). ; Proceedings of the 13th International Conference on Natural Language Generation, INLG 2020 ; Conference date: 15-12-2020 Through 18-12-2020",

year = "2020",

month = dec,

language = "English",

pages = "35--40",

url = "https://www.inlg2020.org/",

}

TY - CONF

T1 - Studying the Impact of Filling Information Gaps on the Output Quality of Neural Data-to-Text

AU - Thomson, Craig Alexander

AU - Zhao, Zhijie

AU - Sripada, Somayajulu Gowri

N1 - Conference code: 13

PY - 2020/12

Y1 - 2020/12

N2 - It is unfair to expect neural data-to-text to produce high quality output when there are gaps between system input data and information contained in the training text. Thomson et al. (2020) identify and narrow information gaps in Rotowire, a popular data-to-text dataset. In this paper, we describe a study which finds that a state-of-the-art neural data-to-text system produces higher quality output, accordingto the information extraction (IE) based metrics, when additional input data is carefully selected from this newly available source. It remains to be shown, however, whether IE metrics used in this study correlate well with humans in judging text quality

AB - It is unfair to expect neural data-to-text to produce high quality output when there are gaps between system input data and information contained in the training text. Thomson et al. (2020) identify and narrow information gaps in Rotowire, a popular data-to-text dataset. In this paper, we describe a study which finds that a state-of-the-art neural data-to-text system produces higher quality output, accordingto the information extraction (IE) based metrics, when additional input data is carefully selected from this newly available source. It remains to be shown, however, whether IE metrics used in this study correlate well with humans in judging text quality

M3 - Unpublished paper

SP - 35

EP - 40

T2 - Proceedings of the 13th International Conference on Natural Language Generation

Y2 - 15 December 2020 through 18 December 2020

ER -